Draft:Cognitive Information Retrieval

Cognitive Information Retrieval

 Introduction :

The word cognitive refers to the process towards knowledge or awareness. By using the terms cognitive science,it links between process of information ,perceptual skill ,resources conceptualized and topic related to the cognitive psychology.

CIR is a multidisciplinary field of study which interact with human computer on the basis of human factors. It is a field which focuses to improving the interaction between users and IR system. CIR is include research from information science to cognitive science. CIR interacting between the user and IR systems by providing the cognitive processes involved in how users search for, interpreate and use information.

CIR is an important part of the cognitive state and difficult to arrange new approaches and development of this new approaches to designed the web and IR systems. CIR is envisioned as complicated human information and human computer interaction process which integrated within an individual’s everyday life context.

 Concept of   Cognitive Information Retrieval 

Cognitive information retrieval  can be defined as the study and implementation of IR systems. This systems aim to bridge the gap between the user’s intent and the information retrieved. This process ensuring that the process is not only efficient but also intuitive and aligned with how user  think and process information.

Key concept

Relevance

Cognitive and interactive

Poly representation approaches

Relevance

The relevance concept currently forms  the basis of users provide feedback on the relevance of retrieved documents,  which is used to improve the relevance of future search result -as opposed to the initially expressed - information need for the user ,represented in the interaction by the user’s query. The search results are tailored  to match the specific context  in which the search is conducted ,such as the user’s task ,location and previous interaction. CIR systems adapt to user’s individual cognitive styles,preferences  and past interactions to deliver more personalized and relevant search results. This view of relevance, Ruthven believes -  with a single, objective reality to relevance based on topic  - has been a major obstacle to the development of more naturalistic method and systems.

Cognitive and interactive

The cognitive view of a user is to interact with the IR systems. In a cognitive perspective ,  information need as a conceptual basis for the user and their interaction is problematic.

Emphasizing continuous interaction between users and systems to refine such queries and adapt to user feedback. Cole, Beheshti , Leide and large then specify the terms and definitions of the user systems interaction they wish to examine in detail. To the end , these interaction are represented as a interacting States: the users task and the users cognitive state ,  the various need State for each task , that may arise during the user’s interaction with the IR system.

Poly representationl approaches

Poly representation approaches typically refers to representing information or data in multiple formats. In  various field such as in NLP, computer vision and multimedia retrieval, this approaches involves encoding data in different representations ,such as text, images ,audio or other modalities.

Burger Larsen and Peter Ingwetsen describe a user and document representation methodology that gives a IR system with a fulller picture of user and the document set than current simple request based systems allow ,  based on the principle of poly representation and approaches.

 History and evolution: 

The cognitive approach in information science and IR research started at the end of the 1970's. The early of 1970's cognitive approaches was influenced by discussions of cognitive science. Cognitive approache’s reviews and discussions to IR during 1980’s  can be found in Belkins overview (1990), Ellie's (1989,1992), critical essays, and Ingwetsen's (1992)book.

The development of the cognitive approaches to IR into two distinct  periods. In the first period 1977 -1991, cognitive IR build on Demey's (1977-1980) original thesis.

In the second period 1992-2000, Demey's core evolutionary view consisting four stages through which thinking on information processing has developed.

The history and evolution of CIR span several decades  and have been shaped by advencement in cognitive science, IR technologies  and uses centered  design principles.

1950-1970 :

During the mid 20th century, pioneers such as Robert Taylor and Calvin  Mooers laid the foundation for understanding human information behaviour.

In this studies focused on how individuals perceive, process and utilise information sitting the stage for integrating cognitive principles into IR.

1980-90’s

In this time researchers starts to developing cognitive models to describe and predict user behaviour during IR systems. Notable models include the cognitive models of IR by Belkin and Vickery, which emphasizes the role of cognitive processes in search behavior.

Late 1990’s-Early 2000 :

In this time witnessed a shift towards user centred  design in IR. Emphasis  was placed on designing systems that aligned with users cognitive processes ,preferences and needs to enhance usability and effectiveness.

2000’s :

In this time saw significant and advancements  in interactive IR, allowing for real time feedback and adoption based on user’s cognitive.

2010’s :

In 2010 there was a growing focus on personalized in IR.Advances in machine learning and data analysis enabled the development of personalized retrieval systems that adapt to user’s cognitive styles  and preferences.

2010-2020:

More recently, there has been a trend towards multimodal retrieval, integrating multiple modalities such as text, images, and audio.Multimodal approaches enable more comprehensive retrieval of diverse types of information, aligning with users' cognitive preferences and abilities.

Throughout its evolution there has been an emphasis on evoluting and validating CIR  system .Researchers have developed Matrics and methods to assess the effectiveness usability and user satisfaction of CIR systems in real world.

 Proponents of Cognitive Information Retrieval :

Peter Ingwerse

Peter Ingwerse propose a cognitive model, this model emphasizes the importance of understanding user mental models. He proposed the interaction between the user and IR system. The components   of this models are user's manual cognitive space, system's cognitive space and Interaction process.(Ingwerse, 1996)

Marcia J. Bates

Marcia J. Bates  describe a non linear and evolving approach to IR. He describe that the user can change their minds and often modify their queries. ( Bates, 1989)

DEMEY’S (1977-1980)

“That any processing of information weather perceptual or symbolic is mediated by a system of categories or concepts which for the information processing device r a model of his world.”

Peter  Ingwerse and kalervo Jarvelin

They proposed the stratified model of IR. This models describes the various interconnected stage, which are from cognitive process to Socio- organisational contexts. The components are Cognitive level, information object level, Socio- organisational contexts.(Vakkari & Jarvelin,2006)

Peter Pirolli and Stuart Card

They proposed Information Foraging Theory. This theory describe that how user's search for and process information, aiming to maximize the value of information obtains relative to the cost of seeking it.(Pirolli & card , 1999)

Belkin

Belkin along with his colleagues, developed the CMIR in 1980’s. His work identifies  the importances  of user’s  cognitive process and impact on IR task.

Tulving Concept

The episodic memory model in IR is inspired by Tulving Concept. This model describe that IR systems should support retrieval based on episodic memory cues. ( Tulving, 1983)

Reference :

1. https://www.rroij.com/open-access/a-review-of-the-cognitive-information-retrieval-concept-process-and-techniques-46-50.php?aid=38180

2. https://en.m.wikipedia.org/wiki/Cognitive_models_of_information_retrieval

3. https://peteringwersen.info/publications/2330_ingwersen_arist34-finalprint.pdf

4. https://www.researchgate.net/publication/259703580_Towards_a_cognitive_theory_of_IR

 Cognitive information retrieval process 

Cognitive information retrieval concept leads to the cognitive information retrieval process, which involves some important criteria like multitasking and human interactive behaviour (HIB) framework.

Now let’s discuss the multitasking process in cognitive information retrieval.

Some famous notable scholars of information, science, and information retrieval Amanda spink and Charles cole analyses about users, multitasking information behaviour while the users are interacting with information retrieval systems.

The current information Retrieval system design for one search task at a time like the users input the queries to find the needed documents for a single topic for better understanding.

When we user searches anything in Google search engine. It gives results about just the topic.

Multitasking

Next, we need to understand about multitasking behaviour which is natural by nature. Users often search for multiple topics during a searching session, they might start with a single topic and later they might switch to other topics according to the need or mood, multitasking is a common and natural behaviour.

But multitasking may be problematic for users as the search engines Are not very good at handling multiple searches at the same time. Information scientists studying. How people change their opinion during searching in online mode..

In cognitive perspective, when users use search engines, their thoughts and needs affect how they search. Search engines should show information that fits with what people are thinking and wanting at that moment.

Multitasking can be better understood with users searching techniques. Let’s try to understand with an example.

If a user is searching for Yoga to get fit then suddenly switch that search to proper diet or can search for equipment for exercise.

2.  Human interactive behaviour (HIB) framework.

The human interactive behaviour framework interacts with electronic information retrieval systems, like search engines or digital libraries.

This interaction has been examined for over 50 years that evolved with advancements in technologies.

Users have been using search engines and online libraries for a long time. The tools have been updating and getting better day by day over the years.

Researchers from various films like library & information science, cognitive science, human factors and computer science are involved in how users use search engines to get information.

Previously, the researchers only now focused more on how people use them.

The researcher is researching on how users interact on information retrieval systems, though it is an important area, still the area underfunded and not widely discussed at the technology conference.

Researchers and combining studies of how users think and search online with how users behave overall when using search engines.

Human interactive behaviour is very crucial for better information Retrieval systems. By developing better tools that fit natural human behaviours, researchers can make a better information retrieval system.

Relevance Feedback

The cognitive information retrieval process evolved into techniques that are relevant feedback, knowledge visualisation and training framework.

The relevant feedback process refers to the techniques that are used to understand better and improve the interaction between users and the system.

The main objective is to gather implicit feedback from users, behaviour and preferences when interacting with the system.

This feedback is then used to enhance the system's ability to understand and serve the result more naturally to the users.

The process works, when users search anything in a search engine, then the search engine tries its best to show the results, but sometimes it fails.

Relevance feedback is a way for the search engine to learn from the user. it doesn’t require users' opinion to explicitly say what’s good or bad, rather it watches human users interact with results.

The work of Diane Kelly identified Fine Way User typically interacts with search results.


 * 1) Examine: briefly check something out to see if the result is relevant.
 * 2) Retain: saving something that is useful for later.
 * 3) Reference : it is like vice versa of the retain process that is going back to something that is already saved because it was helpful.
 * 4) Annotate: making notes on something(if the tool allows) to remember important parts.
 * 5) Create : using the result information to create further, something new like a report, or presentations may be.

The relevance feedback process helps search engines become better at understanding users' needs.

1.  Knowledge Domain Visualisation (KDV).

The process includes knowledge domain visualisation.

Peter Hook and katy B’orner explores visual techniques to make knowledge domain maps (like topic mind maps) more user-friendly for learning.

If we take an example, like a complex topic of history. A knowledge domain visualisation is like a mind map that shows the different areas of history( like ancient, mediaeval, modern, et cetera) and how they connect.

This helps users, especially those who are new to the topic, understand the structure and find information more easily. Knowledge of visualisations can be part of search engines or digital libraries.

2.  Training Framework

Now another process is training framework:

Wendy lucas and heikki Topi describe how to train users to search for information more effectively in information retrieval systems.

They proposed five stages so that the search training should be effective for seeking information.


 * 1) Articulating the need: this stage tells about clearly Defining the information that the user is looking for.
 * 2) Conceptualising the query: turning the information needed into a general search query.
 * 3) Formulating for the system: adapting the query, according to the specific information retrieval system that is in use.
 * 4) Entering the query: type the search terms into the search bar.
 * 5) Understanding results: making sense of the information the system returns.

It is important because each stage requires difficult skills training at. One might focus on helping users define their needs, while stage three might teach users about how to use features specific to a particular search engine.

A common mistake users make (like forgetting Boolean operators) can be identified and addressed in training to improve overall search effectiveness.

References
 * 1) Bhatia, P. K., Choudhary, C., Mehtrotra, D., & Wahid, A. (2013). A review of the cognitive information retrieval concept, process and techniques. Journal of Global Research in Computer Science, 4(3), 46-50.
 * 2) Smeaton, A. F. (1999). Using NLP or NLP resources for information retrieval tasks. In Natural language information retrieval (pp. 99-111). Dordrecht: Springer Netherlands.

 The Holistic Cognitive View 

The holistic cognitive theory for IR originated 1990-1992 and became more profoundly analyzed in (Ingwersen, 1996) leading to an increasing weight of empirical studies based on hypotheses derived from that theory (Ingwersen and Järvelin, 2005). It replaced a more individualistic perspective of the cognitive view in (interactive) IR dominant from mid-1970.

A holistic cognitive view in library science encompasses a comprehensive understanding of how individuals interact with information and libraries, taking into account psychological, sociocultural, and technological factors. Here are some key components:

Information Behavior: Recognizing that people's information-seeking behaviors are influenced by various cognitive processes, such as perception, attention, memory, and decision-making. This perspective considers how individuals acquire, process, and utilize information in different contexts.

User Experience (UX) Design: Emphasizing the importance of designing library services, systems, and spaces that are user-centered and intuitive. This involves understanding users' cognitive processes, preferences, and needs to create seamless and engaging experiences.

Information Literacy: Viewing information literacy as more than just a set of skills, but also as a cognitive process involving critical thinking, problem-solving, and communication. A holistic approach to information literacy includes considering learners' cognitive development and metacognitive abilities.

Social and Cultural Contexts: Acknowledging that cognitive processes are shaped by social and cultural factors. This perspective explores how cultural norms, societal values, and community practices influence information behavior and learning in library settings.

Technology Integration: Recognizing the impact of technology on cognitive processes and information behavior. This involves leveraging digital tools and platforms to enhance access to information, facilitate collaboration, and support lifelong learning.

Human-Computer Interaction (HCI): Applying principles from HCI to design user interfaces and information systems that are cognitively efficient and user-friendly. This includes considerations of cognitive load, mental models, and information organization.

By adopting a holistic cognitive view, library professionals can better understand the complex interactions between individuals, information, and technology, leading to more effective library services and resources that meet the diverse needs of users.

The reasons for initiating the cognitive research at the beginning of the 1990s are several.

The SCHAMBER ET AL. article on situational relevance in 1990 inspires an increasing amount of hitherto unanswered or ignored questions about relevance phenomena in fora outside the algorithmic IR domain. For instance, HARTER proposes in 1992 the conception of psychological relevance and its relationship to information need formation. With respect to the concept of information need, he challenges the ASK hypothesis proposed a decade earlier by BELKIN ET AL. (1982a). However, he misinterprets the central point of the hypothesis, that is, that the system design in the paper by Belkin et al. is based on the idea of a constantly changing ASK, exactly by virtue of the interaction process over IR session time. Essentially, Harter (p. 610) restates the Belkin et al. view and adds nothing new to the information need and ASK discussion. What is important is the quite comprehensive understanding of the relevance issue seen as a dynamic and complex phenomenon.

This discussion of relevance complexity is in direct contrast to the view on relevance and interaction taken in the large-scale mainstream TREC experiments that start in 1991-1992 (HARMAN). In TREC, continuing the empirical Cranfield tradition, relevance is regarded as a binary, topical, and stable, manifestation. TREC forces IR research to reconsider the nature of IR and IR theory. IR theory is not simply algorithmic solutions to technical problems in settings without realism, nor just socio-psychological theories of user behavior in more realistic settings. Both mainstream, cognitive and user-centered theorists recognize the fragmented often black-box like state of affairs at this point in time, and some serious attempts are made during the ensuing years to clarify and understand the situation.

ROBERTSON & HANCOCK-BEAULIEU point to three recent revolutions in IR that, in their opinion, are crucial to understand in order to proceed toward a more holistic theory of IR: the cognitive, the relevance, and the interactive revolutions. The cognitive and relevance revolutions in empirical IR require realism with reference to the processes of information need development and human relevance assessments. This means that an information need ought to be treated as a potentially dynamic concept, and that the multidimensional and similarly dynamic nature of relevance should be taken into account. Relevance ought to be judged against the information need situation, not the query or request, by the person who owns the information need or problem situation. The interactive revolution points to the fact that (even experimental) IR systems have become increasingly interactive, due to actual applications of dynamic relevance feedback and query- modification techniques by users over IR session time. Thus, experimental or evaluative IR settings, as well as theory, have to incorporate this realism that would incorporate the context or situation surrounding the IR activity, i.e., seeking processes. At the same time, experiments must maintain a degree of control. The question is, what do IR researchers wish to observe, analyze, measure, or make theories about? The revolutions can thus be seen as the real challenge to the IR community.

DE MEY’s (1980; 1982) original view of cognition in contextual social interaction as well as his stages of information processing, their effect on the concepts of information and information need formation, and their association with IR research are not fully explored. An attempt is thus made by INGWERSEN in 1992 and 1996 to discuss in detail the state of IR theory and research from an interactive perspective, hereby following up on the work by BELKIN & VICKERY.

By contrasting the usual relevance notion of topicality with the concept of situational relevance, SCHAMBER ET AL. also stress the importance of context in IR. Situational relevance derives from P. WILSON’s original concept in 1973. Context may come from the information objects or knowledge sources in systems, but may also be influenced by the actual information seeking situation adhering to a domain. Situational also implies a series of dynamic cognitive states in the mind of the user during an IR session. Essentially, this means that if relevance evaluation is dynamic, the corresponding information need is dynamic as well. During an IR session, an information object may thus be topically relevant in the sense of the TREC experiments (HARMAN), but may not be useful to the situation the end user is facing at that time. Obviously, only the user can assess this type of relevance.

There is some similarity between situational relevance and LURIA’s situational classification of objects. For example, one can easily detect situational assessments in the empirical studies of user-intermediary-system interactions. The user’s current perception of the situation, rather than the user per se, becomes the focal point for IR interaction. The intentionality of the user (in the sense presented by SEARLE, i.e., the user’s goal or purpose) and background knowledge at a given time can be seen as the crucial components in IR interaction. IR implies a continuous process of interpretation and cognition. In a holistic sense such processes take place both on the user side and on the system side during human-machine interactions. However, DE MEY's (1980; 1982) stages of understanding information processing in a broad sense define the limitations and characteristics involved on both sides in a cognitively asymmetrical way. The shift into a holistic cognitive view implies that the belief that the variety of cognitive structures in IR (e.g., indexing structures, information object structures by authors, or users' cognitive interpretations) can be commonly understood shifts to the acceptance that such structures are inherently different. The goal is to explore and employ the cognitive differences of structures in such a way that IR can be facilitated and improved. At its current level of cognition, i.e., recognition and algorithmic processing following implemented rules, the machine cannot become aware of and understand what the user is looking for, in particular when the user's ASK initially is ill-defined. The machine may, however, be designed to support the user during the interactive process of IR in order to make the user interpret and learn so he/she feeds data back to the machine, which eases its way of further supporting the user. Ultimately, information retrieval in its real sense takes place only in the mind of the information seeker, that is, information is seen as process or as knowledge and cognition, as discussed by BUCKLAND (1991a; 1991b). The road is constantly uphill from the machine point of view.

DIMENSIONS OF THE HOLISTIC VIEW

During this second period the understanding of information as a situational phenomenon is quite clear. A direct consequence of the situational approach and the enhanced cognitive model is a shift away from the idea of simply bringing cognitive structures into accord. Instead, the new cognitive IR theory focuses on the explicit application of the cognitive structures of different origin involved in IR interaction. The objective is to point to the potential value of matching the multidimensional variety of representations inherent in or extracted/interpreted from information objects and from the cognitive space of a user in a social and situational context. The work task and its perception by a user is regarded as just as valuable as the information need in the form of requests (INGWERSEN, 1996).

Poly-representation of Information Objects

Cognitive IR theory favors all kinds of variations in information structures, and particularly favors retrieval overlaps between such variations. The assumption is that the more disparate the structures in cognitive origin, logic, functionality, and time, the smaller the overlap and the better and probably more relevant the retrieval outcome (INGWERSEN, 1996). The concepts of cognitive retrieval overlaps, data fusion (BELKIN ET AL., 1995b), and request fusion are essential elements of a theory framed by the cognitive perspective. For example, SARACEVIC & KANTOR point to retrieval overlaps and their potential for improved retrieval outcome in a combined information seeking and retrieval user study.

The overlaps are based on a principle of multiple evidence or poly-representation. The principle originates from the arguments proposed by TURTLE & CROFT in 1990 with regard to their generic inference network proposal. The network implies different ways of referring to the same concept and of linking different concepts in the form of a conceptual net thrown over the underlying information objects. Following INGWERSEN (1992, p. 201), the principle implies analog representations in a variety of different forms of one information object, or of an information requirement. Fundamentally the representations are forms of the different cognitive structures noted above.

Representations of different cognitive origin and type are generated via a variety of well-known retrieval and (automatic) indexing methods applied to IR systems. Different indexing methods, for example, natural language versus controlled vocabulary, applied to the same text collection retrieve different sets of text or passages for the same query. A passage can equal a paragraph, a section, or a figure in a full-text object. In online bibliographic databases, human indexing with a small number of controlled terms per document results in a representation based on human interpretation and domain expertise of the entire full-text object. As such, this leads to a heavy reduction in access possibilities, although new facets of potential informative value may have been added by the indexer. The automatic natural-language processing (NLP) approach provides many more access points, but generated by the author. The author’s terms in section headings or titles are assumed to have greater weight than similar words in the text body. Hence, the writing style in scientific communication influences the NLP result differently due to its domain dependence. The NLP result can be automatically filtered through a domain specific thesaurus of controlled terms. Thesauri are actually interpretations of a domain by experts different from the indexers (INGWERSEN, 1996).

Several empirical investigations carried out in the operational Boolean environment demonstrate clearly that the combination or the overlap of controlled index terms and natural-language representations yields better retrieval results than the two separately (KATZER ET AL.; LANCASTER; TENOPIR). The more variety in cognitive origin of representation, the more different the results and narrower the overlaps. Because of individual variance in interpretation, human indexing results in retrieved sets rather different from those based on automatic indexing. INGWERSEN (1996) exemplifies how the mixing of author-generated title and abstract terms with indexer-generated terms can be applied following a Boolean quorum logic combined with the cognitive principle of poly-representation or multiple evidence in traditional online bibliographic databases. AHLGREN (1998; 1999) proposes how to improve and logically simplify this cognitive online approach.

One of very few investigations of the overlap in retrieval of journal articles using term indexing and citation indexing was carried out by MCCAIN. As expected, for the same queries in Medline and the citation databases, the two very different cognitive structures of indexers and citing authors yielded an overlap of only 11% on average. PAO’s (1993; 1994) similar investigations of the overlap between indexing and citation analysis went further by clearly demonstrating that up to 90% of retrieved documents in the narrow overlap set were judged relevant by experts. HARTER ET AL. investigated two confirmatory representative structures associated with scientific communication—those of citing and cited authors—and observed their semantic relationships or overlaps.

SWANSON’s 1986 methodology for examination of scientific communication patterns to discover hitherto unknown connections between two medically remote research communities belongs to the same kind of multiple evidence approach. There is a small but increasing interest in these approaches to retrieval based on different types of citation indexing. When invented three decades ago, citation indexing was seen as an alternative to text retrieval (GARFIELD). Now in light of current cognitive theory and information technology, they appear to offer far more in combination than anticipated.

With respect to overlaps between sets of texts retrieved via different best-match (and Boolean) techniques or search engines for the same user statement, the picture is identical. From a cognitive perspective, this evidence of variety is obvious. This is the idea behind data fusion as discussed by KANTOR and first put into practice in the I3 R system developed by CROFT & THOMPSON.

An interesting line of research is proposed by VAN RIJSBERGEN & LALMAS and LALMAS & RUTHVEN. Van Rijsbergen & Lalmas suggest the application of uncertainty logics, including abduction, because uncertainty and unpredictability are fundamental obstacles to effective access to information in IR. They go on to propose the application of the Dempster-Schafer theory of evidence as a logical tool in IR. The proposal is carried on by Lalmas & Ruthven. In many ways their approach seems associated with or complementary to cognitive theory, primarily with regard to the use of uncertainty and multiple evidence or polyrepresentation of information objects. A comprehensive cognitive IR theory should not only attempt to bridge or associate with prevailing theories in mainstream research, but it ought also to be able to explain some fundamental problems that occur in such theories. One example is the well-known but unrealistic assumptions of term and relevance-assessment independence concerning probabilistic and vector-space techniques (ELLIS, 1996). The term-independence assumption implies that each term or feature in an information object is independent of other features. Terms are the entities on which probabilistic or vector calculations are made, and they seem to yield a better performance result on average than the use of concepts or composite terms or phrases. Similarly, each relevance assessment of a ranked list of objects is assumed to be independent, or not to influence other assessments. The first assumption is in fact realistic in a cognitive sense if we assume that authors actually write meaningful texts and the query is rich. In such cases (INGWERSEN, 1996), when a text passage is retrieved containing a vast number of independent query terms, the probability that we reach a meaningful text entity is high and correct topicality is assured. Clearly, we have not necessarily retrieved information in a cognitive sense, but conceivably a meaning quite identical to the query.

DRETSKE's semantic information theory comes to mind, in which information (seen as the text itself, as opposed to the cognitive information conception) leads to meaning, that is, to making sense. In addition, although phrases are more concise and precise features, they are more semantically closed than single terms. In a cognitive sense, the bigger the context, the more disambiguation at the semantic linguistic level (Figure 1), and the fewer the paths into information space of possibly useful nature to the user. Hence, phrases limit recall. The relevance-independence assumption should be seen in the light of the Cranfield-like experiments in which titles or short abstracts were presented to users. In that case the assumption does not necessarily hold because each assessment will be as fast and possible to remember as the ensuing one. This depends, of course, on the presentation mode of objects. However, if full-text objects are assessed, the independence assumption may very well hold due to cognitive overload during assessments.

 Cognitive Space and IR Interactions 

Two particularly important structures in the cognitive space of the user are the perceptions of the work task and the information need that drive the user’s seeking behavior. Various researchers are finding ways to observe and understand the relationship of work task to information need with the goal of designing more responsive IR systems that employ mechanisms such as feedback and query expansion.

Work task. As part of the situational context surrounding the world model of the user (Figure 1) are the work tasks imposed by the social-organizational environment and perceived by the user by means of his or her current cognitive state as an interest, problem, or task to be pursued (INGWERSEN, 1996). This act of perception can be seen as a dominant component of the problematic situation conceived in the ASK hypothesis (BELKIN ET AL., 1982a), or the component regarded as the cause for information need development. In a cognitive sense the user's perception of a work task is likely to be more stable over IR session time than the corresponding dynamic information need. The perception of work task is thus appropriate to utilize, as it may provide the context necessary for the system to retrieve relevant information, i.e., information useful to that user in carrying out the work task. This conception of context and information use is associated with WILSON (1981). In this respect we may see how decade-old conceptions, proposals and observations often are resurfacing and tuned into new shapes due to the ongoing process of research in the field.

Several extensive studies of the influence of work task on IR have been made. BYSTRÖM & JÄRVELIN, in a study based partly on the cognitive view and partly on sense-making theory, investigated the complexity of work tasks and how it affects not only IR but also information seeking and use in decision making. Their results show that, in the case of complex and ad hoc work tasks, both the information need and corresponding problem perception are extremely weak or nonexistent. The user can describe only the perceived work task itself. Evidently, any IR design and/or evaluation ought to take that factor into consideration. Researchers should investigate the socio-organizational environment and domains with respect to characteristics of work tasks and preferences as well as their associated problem and need manifestations. This type of domain analysis can be carried out by means of common social science data collection and analysis methods, such as process analysis. A detailed model for work-task analyses is provided by RASMUSSEN ET AL. (p. 206) in their empirically based cognitive ecological framework, which covers both design and evaluation issues of information systems and is both analytical and empirical.

In evaluation of IR systems, BORLUND & INGWERSEN (1997) created simulated work-task situations, or cover stories, that were employed by test users to generate their own information needs. Although the study was small in scale, the results indicate that IR systems can be tested applying simulated task frames. Currently, the technique of applying real information needs generated by test users in response to simulated work tasks is being studied in large-scale IR system evaluation experiments involving full-text and best-match algorithms including query modification (BORLUND; BORLUND & INGWERSEN, 1999). ALLEN (1996a, 1996b) and HERSH ET AL. use the work-task approach in a more pragmatic and user-centered study, while JOSE ET AL. rely on the Borlund technique of simulated tasks and information situations in an evaluation study of image retrieval. REID supports Borlund’s technique in promoting a task-oriented approach to IR evaluation in general. It should be noted that a task can be understood in two ways in IR: (1) as a work task that originates external to the user from an information and domain situation and (2) as one or several different retrieval or search tasks of specific conceptual or retrieval nature to be performed in connection with searching for information. Hersh et al. study the task phenomenon in its latter capacity as does MARCHIONINI, who regards information seeking as a process in which the search task consists of a series of actions in pursuit of an aim. The aim is identical to the work task or interest. Search tasks are also (re)investigated in relation to online searching, i.e., searching virtual environments, such as digital libraries or the World Wide Web. Various approaches to investigating cognitive style are recently discussed. FORD introduces the use of neural networks to handle fuzzy navigational behavior, while PALMQUIST & KIM observe search performance in terms of time spent and number of nodes traversed, and S. PARK looks into the role of integrated interaction and user control of the distributed environment including database selection.

The work task and its effect on the user’s perception constitutes a valuable dimension of the cognitive space of the user, in line with problem and information need perceptions. Conceivably, this multidimensionality can be further exploited in building request models to extract descriptions of various user perceptions.

Feedback and query expansion. SPINK ET AL. (1998a) conducted a large-scale empirical investigation of elicitations of information from users by search intermediaries in standard online IR situations. By means of 40 mediated IR interactions and more than 1500 elicitations, the authors establish a categorization of purpose and strategies for mediated questions from users with real needs and monitor the transition sequences from one type of question to another. Their conclusions are compared to previous interaction studies and models (BELKIN ET AL., 1987; INGWERSEN, 1982, 1996; SARACEVIC, 1996a; SARACEVIC ET AL., 1990) and stress in a very detailed manner the importance of extracting information from the user on search terms, domain knowledge level, previous information seeking experience, and search knowledge. SPINK & SARACEVIC (1997) found that search terms evaluated from systems feedback and later applied to query expansion were found to be highly productive, as was the interactive IR session in teams of users.

Feedback from IR systems is thus a fundamental element both in standard online retrieval and in relation to best-match IR, for instance, in TREC. Based on the cognitive approach and cybernetics theory, SPINK analyzes three different feedback frameworks applied in information science research. She suggests enhancing the feedback concept within the cognitive understanding of information, thus illuminating the information seeking and retrieval context. In an ARIST chapter, SPINK & LOSEE present a comprehensive overview of feedback issues in IR. SPINK & SARACEVIC (1998) provide a smaller but highly structured review related to human computer interaction in IR.

Manual query modification during experimentation, for example, in the OKAPI experiments (HANCOCKBEAULIEU ET AL.), combined with relevance feedback, demonstrates highly interesting results from a cognitive view. It provides the basis for improved cognition and expression by the user of the underlying problem or work task and, possibly, of the actual need for information, by forcing the user to interpret the search outcome (SPINK ET AL., 1998a). This outcome does not have to be monolithic, that is, one simple ranked list, but may also contain pointers to several conceivable routes into information space, for example, hypertext links, class names, condensed or structured lists of concepts, and analogous means of conceptual feedback. The structures of monolithic term lists have been empirically evaluated by EFTHIMIADIS (1995) from a cognitive and user-centered view. Simple frequency-ranked term lists seem less valuable than lists ranked with the “best” search terms first.

The results of some investigations disagree on the effect of query modification on retrieval outcome. BELKIN ET AL. (1996a) maintain that interactive query modification adds to the performance of the total IR system, measured as recall and precision in interactive TREC experiments. In contrast, HANCOCK-BEAULIEU ET AL. claim that manual query expansion in various forms does not increase overall performance but rather decreases it. The latter result has been confirmed by MAGENNIS & VAN RIJSBERGEN. However, the investigations on this matter are very few and, due to the experimental settings, difficult to compare. We have indications that successful use of ranked retrieval systems depends heavily on the users' mental models of how such systems operate, including functionality like relevance feedback and query modification (BELKIN ET AL., 1996b).

It should be noted that equal treatment, accumulation, or fusion of request versions (or relevance feedback) over IR session time, as done in noninteractive best-match algorithms, is incompatible with a cognitive theory for information transfer. Because cognitive theory assumes that dynamic interpretations, learning and cognition take place during information interaction, the latest versions of requests should be given higher priority or weighting than previous versions in interactive best-match algorithms. This does not make objects redundant that previously have been judged relevant and still are seen as useful by the user. The systems design should obviously allow for a return path or save functions, consistent with the observations of relevance behavior by FLORANCE & MARCHIONINI. CAMPBELL and CAMPBELL & VAN RIJSBERGEN propose an application of a probabilistic model, called the Ostensive Model, that reduces the weights given to previous relevance feedback results, thus conforming to the realistic concept of information need and situational variation during IR. The implemented system is intended to work without query reformulations by the user, a conception similar to that of the THOMAS system (ODDY, 1977b); instead it displays image representations of the information objects in the form of clusters of candidate objects generated by the

highest-ranked objects based on the previous relevance feedback actions with falling probabilistic weights. In the effort to reduce clutter, the candidates are only shown around the current object chosen by the user. The path of previously assessed clusters is the sequence of objects from the starting point to the current object. The user is thus free to select any object at any time as the new current object (CAMPBELL).

Relevance and Evaluation

The TREC and other interactive investigations and experiments demonstrate the problematic issues concerned with the concept of relevance and the evaluation methods generally applied in IR. Until recently a basic drawback of the cognitive approach has been a lack of discussion directed toward these issues. Yet, the number of journal articles on relevance alone has increased tremendously since the SCHAMBER ET AL. article in 1990: in that year 9 articles were published and in 1996-1997 the trend peaked with more than 50 articles per year. Schamber reviews the issue of relevance in ARIST (SCHAMBER, 1994) and has herself contributed to the empirical study of relevance assessment criteria from a cognitive approach (SCHAMBER, 1991), along with others, including BARRY, BARRY & SCHAMBER, BRUCE, and T. K. PARK. A thorough discussion of evaluation of IR systems can be found in the ARIST 32 chapter by HARTER & HERT.

The relevance experiments and investigations, including those associated with TREC, led SARACEVIC (1996b) to produce the most comprehensive model of relevance types and IR interaction. The model is seen as an alternative to but strongly associated with the cognitive models proposed by INGWERSEN (1992; 1996) and BELKIN ET AL. (1995a). Saracevic’s model is at least two-dimensional. One dimension is occupied by three communication levels.

The (1) processing level corresponds roughly to SMEATON’s morpho-lexical and syntactic linguistic levels, and the (2) interactive and (3) cognitive levels to Smeaton’s semantic and pragmatic levels.

The other dimension consists of five increasingly subjective types of relevance: from (1) algorithmic, which is similar to ranked machine output, e.g., in TREC; through (2) topicality; (3) pertinence to information needs; and (4) situational relevance. Situational relevance corresponds to the work-task situation.

Saracevic also introduces (5) an emotional/intentional type of relevance that can be seen as a socio-cognitive assessment category referring to the domain and its collective preferences (COSIJN & INGWERSEN). Later SPINK ET AL. (1998b) add two new dimensions to the model: time and relevance scaling. BORLUND & INGWERSEN (1998) propose the application of scaling during IR experimentation and two new performance measures built on the Saracevic model: relative relevance, which compares different types of relevance assigned by users to the same ranked output; and ranked half-life measures, which measure the capacity of the system to rank the most relevant object as high on the list as possible. From a more logical and structured stand, MIZZARO (1996; 1997) analyzes the variety of conceptualizations of relevance, including cognitive contributions in IR.

References

INGWERSEN, PETER. 1996. Cognitive Perspectives of Information Retrieval Interaction: Elements of a Cognitive IR Theory. Journal of Documentation. 1996 March; 52(1): 3-50. ISSN: 0022-0418; CODEN: JDOCAS.

INGWERSEN, PETER; PEJTERSEN, ANNELISE MARK. 1986. User Requirements: Empirical Research and Information Systems Design. In: Ingwersen, Peter; Kajberg, Leif; Pejtersen, Annelise Mark, eds. Information Technology and Information Use. London, UK: Taylor Graham; 1986. 111-124. ISBN: 0-947568-06-9.

INGWERSEN, PETER; WORMELL, IRENE. 1988. Means to Improved Subject Access and Representation in Modern Information Retrieval. Libri. 1988; 38(2): 94-119. ISSN: 0024-2667.

KUHLTHAU, CAROL COLLIER. 1991. Inside the Search Process: Information Seeking from the User's Perspective. Journal of the American Society for Information Science. 1991; 42: 361-371. ISSN: 0002-8231; CODEN: AISJB6.

 Cognitive information retrieval practice and behavior 

The cognitive information retrieval (CIR) is to comprehend and simulate the mental operations users perform when looking for and retrieving information. With an emphasis on cognitive behaviors and strategies, this method takes into account how users perceive, process, and use information.

In information retrieval, the interpretation of information seeking of users perceive is the 1st step. Such as when a user searches with a term ‘picture of an apple’, then if he wants to get a picture of red apple, if he gets the black coloured or green, he will explain it as wrong information as he has no idea of green or black apple. He has the knowledge of red apples not black or green.

The ability to remember and recognize relevant information is crucial for effective information retrieval. For example, if a user needs information on library classification but only searches using the term "classification," they might receive irrelevant results, such as those related to societal classification or knowledge classification. Through past experiences, users learn to refine their searches with more specific terms, like "library classification," to obtain the exact information they need.

The next stage involves users processing and analyzing data. This includes how they conduct searches and gather information. In a computer class, for instance, if a teacher asks students to research "DRM" (Digital Rights Management), searching for just "drm" might yield varied suggestions like "drm in railway," "drm media," "drm in games," "drm software," "drm full form," or "drm radio." Without knowing that "DRM" stands for Digital Rights Management, students might get confused by results such as "Divisional Railway Manager." Understanding the full form and context is essential for retrieving accurate information.

This process highlights how users solve problems and make informed decisions by effectively utilizing retrieved data. Meeting information demands and reaching accurate conclusions require users to apply proper search techniques and contextual knowledge to navigate through diverse results and pinpoint relevant information.

The upper mentioned overall process depends upon the Interactive Search Behavior, in the first step is Query Formulation. This is the way in which users formulate inquiries to convey their information demands. If there is the question on ‘comparative study between hobby and passion’, if users may search the following hobby or passion, hobby vs. passion, difference between the term hobby and passion, comparison between hobby and passion, hobby NOR passion etc. Suppose they may get 50 documents on their search term, after that they have to choose the Relevant documents. This is called Relevance Judgment, by which users evaluate the applicability of documents they have found. It depends on the user's behaviour. If they are critical thinkers they will go through all the documents, and choose what he needs. Some of them take the 1st five documents or so on. This engages the ‘Search Tactics and Strategies’ by which users employ to focus and sift through search results; one example of this is Bates' `berry-picking" model, which delimitation a more fluid and dynamic search procedure. This model says that information seeking more frequently takes the form of "berry picking," or locating knowledge piecemeal via a variety of sources, rather than adhering to a methodical search process. Her idea states that humans generate the remaining information through three forms of conduct: purposeful search, browsing, and monitoring. The majority of information is generated by humans through passive, undirected action. According to Bates, most browsing and focused searches are the result of a sample and selection process that she calls "berry picking," which has developed from conventional forage and mating habits. Information retrieval research was long preoccupied on the system, not the user. The concept was "one query, one use." This indicates that when a user asked an information system a question, researchers assumed the system would provide a response. Naturally, the user or the system might change the query if the answer was not exactly what they were looking for, but if the best response was identified, it was expected that the user would print the data and the search would be concluded (Bates, 2005, pp. 58–59). But nowadays this concept is changed, as user’s questions are indexed for searching, like if a user searches with ‘Kharasrōtā’ in English, in the place of a drought river he may get the answer Kharasrōtār river. According to Peter Morville (2005) in his book Ambient Findability, Bates's berrypicking model—which shows how users' queries and information needs change as they interact with documents and systems—exposed the shortcomings of the traditional information retrieval model and advanced our understanding of information seeking behavior (p. 59). According to Bates (1989, pp. 409–410), "Every new piece of information they encounter gives them new ideas and directions to follow and, consequently, a new conception of the query." As a result, the query itself may alter in addition to the search words for every new page found. According to Bates (p. 410), this is a "evolving search." Similar to how berries are dispersed throughout bushes rather than appearing in bunches, the various pieces of information found at each stage of the constantly evolving search are dispersed across several sources. Consequently, Bates (1989) referred to this process as "bit-at-a-time retrieval" and said that it is "a realistic model of how people go about looking for information." (Page 421)

Adaptability and learning are crucial aspects of how users continuously refine their search tactics based on their past experiences and the results they get. This process allows them to become more efficient in information retrieval practice. For instance, if a user notices that using specific keywords provide better results, they will likely use those terms in future searches. Users learn to use their strategies by using advanced search features, applying filters, or rephrasing their queries to improve the accuracy and relevance of the search feedback. This adaptive learning process is ongoing and dynamic. It involves not just recognizing patterns in successful searches but also understanding and correcting mistakes. For example, if a user frequently encounters irrelevant results, they might experiment with different search engines, utilize Boolean operators, or seek help from online tutorials to enhance their search efficiency. Over time, these learned behaviors become ingrained, making the user more adept at navigating complex information landscapes.

Interpersonal communication is important for defining information needs and enhancing search results. Effective communication between users and intermediaries, such as librarians, can significantly improve the quality of the search process. Librarians and other information professionals have specialized knowledge and skills that can guide users in articulating their information needs more precisely. Through a collaborative dialogue, librarians can help users clarify their queries, suggest relevant search terms, and recommend useful resources or databases. This interaction is especially valuable in complex or unfamiliar subject areas where users may struggle to identify the right keywords or search strategies. By asking probable questions and offering advice, librarians can bridge the gap between the user's initial query and the desired information. This process not only improves the immediate search outcomes but also contributes to the user's long-term learning and adaptability in using Information Retrieval practice and behavior. For Example, if a user’s research topic is on the sculpture of Harappa civilization, if he asks about ancient Indian history, he may not even get the accurate information he needs, but through the interview process or interaction the library professionals identify his need and then he suggests to him the relevant documents. After getting the accurate information the user will search with the term by which he gets the right information. This is the evolution of information retrieval behaviour. Users' ability to learn from their experiences and refine their search tactics, combined with the guidance and expertise of information professionals, enhances the efficiency and effectiveness in the information retrieval practice.

Cognitive information retrieval practice and behavior related to the term ‘Need for cognition’ (NFC). NFC is an individual perception and cognitive ability of the person. Actually this NFC is described as “differences among individuals in their tendency to engage in and enjoy thinking” (Cacioppo and Petty, 1982, p. 116). High levels of NFC positively affect retrieval practice (Weissgerber et al., 2018). While some researchers accept a linear relationship between NFC and cognitive ability (Fleischhauer et al., 2010 and Hill et al., 2016), where Gärtner et al. (2021) denied this relation. They claim NFC is more about the willingness to put in effort and self-control rather than cognitive ability. Users with lower NFC prefer surface retrieval strategies and haven't challenging retrieval behaviour. Persons with low working memory and NFC can't use effective searching strategies. A challenge in education is to find methods that can motivate students with low NFC to exert more cognitive effort. Retrieval practice could be a potential method for this purpose (Gonthier and Roulin, 2020).

Non-invasive brain imaging methods like functional magnetic resonance imaging (fMRI) have been pivotal in understanding how retrieval practice benefits long-term retention. Studies have shown activity differences during retrieval practice, Enhanced brain activity in fronto-parietal regions has also been linked to the positive effects of retrieval practice, regardless of cognitive proficiency (Jonsson et al., 2020). However, it remains unclear if these patterns are consistent among individuals with varying levels of Need for Cognition (NFC).

There has a relationship between Need for Cognition (NFC) and brain activity during retrieval practice, positing that NFC levels might influence the effectiveness of retrieval practice through semantic elaboration (Jonsson et al., 2020; Wiklund-Hörnqvist et al., 2020). Three possibilities are considered: high NFC individuals benefit more due to greater semantic elaboration; low NFC individuals benefit more due to a pronounced difference between passive and active learning; or retrieval practice benefits all individuals regardless of NFC level (Martin and Chao, 2001; Binder and Desai, 2011; Eichenbaum, 2017). It shows behavioral equivalence in retrieval practice translates to neural equivalence ( Martin and Chao, 2001; Binder and Desai, 2011; Eichenbaum, 2017; Roediger and Butler, 2011; Dunlosky et al., 2013; Fazio and Marsh, 2019; Moreira et al., 2019; Agarwal et al., 2021; McDermott, 2021).

Carola Wiklund-Hörnqvist (2022) assured in “Retrieval Practice Is Effective Regardless of Self-Reported Need for Cognition-Behavioral and Brain Imaging Evidence” that most of the students have learnt many words by information retrieval practice. This follow-up to Jonsson et al. (2020) combines functional brain imaging and behavioral data to offer a thorough explanation of how individual variances in NFC affect how well retrieval practice helps learn vocabulary from a foreign language. It illustrates how NFC and retrieval techniques interact with data from functional and behavioral brain imaging. Such as it is got in the research that half of the Swedish words are recalled or memorised by the information retrieval practice.

Hereby some cognitive models will be discussed which help in information retrieval practice and behavior.

Belkin's Anomalous State of Knowledge (ASK) Model posits that information-seeking behavior arises from a gap or anomaly in a person's knowledge. This anomaly drives the need to search for information to resolve the uncertainty or deficiency in their understanding. It emphasizes user centered design, considering user intentions, contextual factors, and continuous feedback to refine search strategies and improve system interactions. Wilson's Cognitive Model emphasizes the role of cognitive processes in information seeking, including the recognition of information needs, the search for information, and the use of information.

Kuhlthau's Information Search Process (ISP), the model outlines the stages users go through during the information search process, including initiation, selection, exploration, formulation, collection, and presentation.

In Advanced Techniques, Wersig and Hennings (1983) explored the use of conceptual mapping to represent both query expressions and document texts. This approach aims to overcome the limitations of traditional Boolean logic by using probabilistic methods to improve search accuracy and relevance. The integration of microcomputers and advanced software can enhance the flexibility and individualization of IR systems, making the search process more intuitive and user centered.

 Cognitive IR Theory 

Cognitive IR theory focuses on the mental processes involved in information seeking and retrieval, emphasizing the user's perspective. This theory considers how users formulate queries, interpret search results, and integrate new information with their existing knowledge. Concepts: Mental Models: Users form mental representations of how IR systems function, which influence their search strategies and interactions. Effective IR systems should align with these mental models to facilitate easier and more intuitive use. Cognitive Load: The mental effort required to perform a search task.IR systems should minimize unnecessary cognitive load to enhance user performance and satisfaction.information Scent: Cues or signals that help users assess the relevance and value of information sources.Strong information scents guide users more effectively to pertinent information(Schamber &Marchionini, 1996). User Context: The user's knowledge, experience, situational context, and information needs.Understanding the user context helps tailor the IR system to meet specific user requirements.

3. 'Cognitive Models in Information Retrieval′ Several cognitive models have been developed to describe and enhance the information retrieval process. These models provide different perspectives on how users interact with IR systems and process information.

3.1. The Cognitive IR Model by Ingwersen The Cognitive IR Model, proposed by Peter Ingwersen, emphasizes the interaction between the user's cognitive state and the IR system. This model highlights the importance of understanding users' mental models and information needs to design more effective IR systems(Ingwersen, 1996). Components: User's Cognitive Space: Includes the user's knowledge, information needs, search strategies, and situational context. System's Cognitive Space: Encompasses the IR system's design, including its indexing, retrieval mechanisms, and presentation of information. Interaction Processes: The dynamic feedback loop between the user and the system, where user inputs influence the system's responses and vice versa. Application in IR: Adaptive Interfaces: Developing IR systems that adapt to users' cognitive states and preferences, enhancing the search experience. Interactive Feedback: Incorporating real-time feedback mechanisms to refine and improve search results. Example: An academic search engine that tailors its results based on a researcher's previous queries and research focus, providing more relevant and personalized suggestions.

3.2. The Berry-Picking Model by Bates The Berry-Picking Model, introduced by Marcia J. Bates, describes a non-linear and evolving approach to information retrieval. It recognizes that users often modify their queries and strategies as they gather information incrementally, akin to picking berries in a field(Bates, 1989). Concepts: Evolving Search Process: Users refine their search queries and strategies based on the information they gather during the search process. Multiple Sources: Users gather information from various sources and formats, integrating them to meet their information needs. Incremental Information Retrieval: Information is collected bit by bit rather than in one comprehensive search session. Application in IR: Flexible Search Interfaces: Designing systems that support iterative searching, allowing users to modify their queries and explore different sources easily. Integrative Tools: Providing tools that aggregate information from multiple sources and formats, facilitating comprehensive information gathering. Example: A library search system that allows users to refine their searches based on previously retrieved documents and incorporates different types of sources, such as articles, books, and multimedia.

3.3. The Stratified Model of Information Retrieval Interaction The Stratified Model of Information Retrieval Interaction, developed by Peter Ingwersen and Kalervo Järvelin, extends the Cognitive IR Model by incorporating multiple layers of interaction. This model recognizes that information retrieval involves various interconnected strata, from cognitive processes to socio-organizational contexts(Vakkari&Järvelin, 2006). Component Cognitive Level: The mental processes of the user, including perception, memory, and decision-making. Information Object Level: The properties of the information objects being retrieved. Indexing and Retrieval Level: The mechanisms and algorithms used by the IR system to index and retrieve information. Interaction Level:The user’s interaction with the IR system. Socio-Organizational Context: The broader social and organizational environment affecting information retrieval. Application in IR: Multifaceted Interface Design: Creating interfaces that address different cognitive and contextual needs of users. Holistic Evaluation: Assessing IR systems not just on retrieval accuracy but also on user satisfaction and contextual relevance. Example: A corporate knowledge management system that not only retrieves documents based on content but also considers the organizational context and user’s role within the company.

3.4. The Information Foraging Theory: Information Foraging Theory, proposed by Peter Pirolli and Stuart Card, is based on the analogy of animals foraging for food. This theory applies ecological principles to understand how users search for and process information, aiming to maximize the value of information obtained relative to the cost of seeking it(Pirolli & Card, 1999). Concepts: Information Patch: A cluster of related information resources. Information Scent: Cues that help users predict the value of the information they will find. Optimal Foraging: The balance between the effort spent on seeking information and the value of the information obtained. Application in IR: Scent-Enhanced Interfaces:Designing search interfaces that provide strong information scents, guiding users towards valuable resources. Efficient Navigation: Creating tools that minimize the effort required to find relevant information. Example: A website design that includes clear headings, summaries, and keywords to help users quickly identify relevant sections and navigate efficiently. 4. The Episodic Memory Model The Episodic Memory Model in IR is inspired by Tulving's concept of episodic memory, which involves the storage and retrieval of specific events and experiences. This model suggests that IR systems should support retrieval based on episodic memory cues, such as contextual or situational details from users' past interactions(Tulving,1983). Concepts: Episodic Memory: Memory of specific events, including the context and associated emotions. Contextual Retrieval: Using contextual information to aid in retrieving relevant information. Application in IR:Implementing search systems that allow users to retrieve information based on context (e.g., time, location, previous searches).Enhancing query refinement by incorporating user-specific episodic cues.

 Implications of Cognitive Information Retrieval Theory and Model  Cognitive Information Retrieval (IR) theory and models significantly impact the design and development of IR systems by considering the cognitive processes of users. These implications are crucial for creating user centered, efficient, and effective IR systems(Norman, 2013). 1.User Centered Design One of the primary implications of cognitive IR theory is the emphasis on user centered design. By understanding users' cognitive processes, designers can create interfaces and systems that better match users' needs and behaviors. 1.1 Reducing Cognitive Load Reducing Cognitive Load: Cognitive IR models highlight the importance of minimizing unnecessary cognitive load to enhance user performance and satisfaction. This can be achieved by simplifying search interfaces, providing clear instructions, and organizing information in an intuitive manner. Example: Google's search interface is a prime example, where simplicity and ease of use are prioritised to reduce cognitive load. 1.2 Enhancing Cognitive Fit Enhancing Cognitive Fit: IR systems should tailor information representation to match users' cognitive styles and task requirements. This involves understanding how users process information and designing systems that support those processes(Pirolli & Card, 1999). Example: Advanced search options in databases like PubMed allow users to filter results based on specific criteria, aligning with the users' need for precise and relevant information. 2. Adaptive and Personalized Systems Adaptive systems leverage cognitive IR models to dynamically adjust to individual users' preferences and needs, providing personalized experiences. 2.1. Personalization Personalization: Cognitive models facilitate the development of personalized IR systems that learn from user interactions to enhance the relevance of search results. By understanding users' past behavior and preferences, these systems can provide tailored recommendations and more accurate results(Brusilovsky & Millán, 2007). Example: Netflix's recommendation system, which uses algorithms to suggest content based on users' viewing history and preferences. 2.2. Context-Aware Systems Context-Aware Systems: Cognitive IR models emphasize the importance of context in information retrieval. Systems that can adapt to the user's context, such as location, time, and specific task requirements, provide more relevant and useful information(“The Turn,” 2005). Example: Mobile search engines that adjust search results based on the user's current location, providing contextually relevant information. 3. Interactive Search Processes Cognitive models recognize that information retrieval is often an iterative and evolving process. This has led to the development of IR systems that support dynamic and flexible search behaviors. 3.1. Interactive Feedback Interactive Feedback: Providing real-time feedback and suggestions during the search process helps users refine their queries and improve search outcomes. This can include auto-suggestions, query refinement options, and relevance feedback(Ingwersen, 1996). Example: Amazon's search bar provides real-time suggestions as users type, helping them refine their search queries.

 Advantage of cognitive information retrieval 

Cognitive information retrieval (CIR) has several advantages in human cognitive processes. These advantages promote the effectiveness and user satisfaction in CIR systems.

1. User-Centered Design: CIR systems give users' cognitive processes and information needs top priority, which makes for better user-friendly and efficient interfaces. For instance, Belkin et al. (1992) point out that cognitive IR systems are made to comprehend and react to users' mental models, which enhances the naturalness and usability of the search process.

2. Increased Relevance and Personalization: CIR systems offer more relevant and customized search results by taking user context, intent, and behavior into account. According to Spink and Cole (2006), these systems have the ability to dynamically adjust to the preferences of each unique user, increasing the relevance and happiness of information retrieval.

3. Improved Knowledge of User Needs: CIR makes use of advanced user modeling strategies to more precisely predict and satisfy users' information requirements. This leads to a better comprehension of user questions and better alignment with their information-seeking objectives, according to Ingwersen (1996).

4. Effective Information Access: CIR process employ machine learning and natural language processing to comprehend intricate inquiries and deliver prompt, precise answers. Marchionini (1995) talks about this efficiency and highlights how cognitive information retrieval systems can ease the process of finding information and lessen the cognitive strain on users.

5.Support for Exploratory Search: By giving users tools for traversing and visualizing complex information spaces, CIR systems support exploratory search. Bates (1989) provides examples of how these systems help users explore several knowledge pathways, which improves the process of discovery and facilitates thorough information retrieval.

These benefits demonstrate how well CIR systems work in delivering user-centered, pertinent, and successful information retrieval experiences. This efficacy is backed by a wealth of actual data as well as theoretical ideas from the information science community.

References

1. Belkin, N. J., Oddy, R. N., & Brooks, H. M. (1992). ASK for information retrieval: Part I. Background and theory. Journal of Documentation, 38(2), 61-71.

2. Spink, A., & Cole, C. (2006). Human information behavior: Integrating diverse approaches and information use. Journal of the American Society for Information Science and Technology, 57(1), 25-35.

3. Ingwersen, P. (1996). Cognitive perspectives of information retrieval interaction: Elements of a cognitive IR theory. Journal of Documentation, 52(1), 3-50.

4. Marchionini, G. (1995). Information seeking in electronic environments. Cambridge University Press.

5. Bates, M. J. (1989). The design of browsing and berrypicking techniques for the online search interface. Online Review, 13(5), 407-424.

 Conclusion 

More than fifty years of important information science discoveries, experiences, investigations, and experimental testing are incorporated into the cognitive approach to IR theory. It is supported by an abundance of empirical data from studies carried out in operational and informetric settings as well as comprehensive information-seeking research. Crucially, this strategy does not contradict the advances in IR made possible by partial matching techniques. Rather, it effortlessly incorporates cognitive insights into standard IR approaches to improve their ability to meet user wants and behaviors. The cognitive method is positioned as a strong and comprehensive framework for comprehending and enhancing information retrieval systems because to this integration of actual data and theoretical frameworks. When it comes to how information systems comprehend and react to customer inquiries, CIR marks a major advancement.

Information systems' ability to comprehend and react to user questions has advanced significantly with the introduction of CIR. Traditional retrieval systems usually rely on simple statistical models and keyword matching, which can sometimes fail to capture the genuine intent of the user and produce less relevant results. In contrast, CIR generates more accurate and context-aware answers by utilizing cutting-edge technologies like machine learning, natural language processing (NLP), and user behavior analysis. Understanding the underlying intent and semantics of user inquiries is CIR's main advantage. This makes it possible for CIR systems to provide outcomes that are more in line with the needs of the user.

For instance, by examining the context and prior user interactions, CIR may discern between a search for "apple" as a fruit and "Apple" as a technology business. Furthermore, CIR systems gain knowledge from user interactions and keep getting better over time. These systems get more and more skilled at processing complicated requests and providing accurate responses thanks to this iterative learning process. This feature is especially helpful in specialist industries where information retrieval accuracy is crucial, like healthcare, legal research, and academic settings. Conclusively, Cognitive Information Retrieval represents a revolutionary advancement in the development of information systems, greatly augmenting the pertinence and precision of search outcomes. CIR systems provide a more efficient and customized search experience by incorporating cutting-edge AI technology that can recognize and anticipate user needs.

References

1. Belkin, N.J., Oddy, R.N., & Brooks, H.M. (1982). ASK for information retrieval: Part I. Background and theory. Journal of Documentation, 38(2), 61-71. https://doi.org/10.1108/eb026722

2. Croft, W.B., Metzler, D., & Strohman, T. (2009). Search Engines: Information Retrieval in Practice. Addison-Wesley.

3. Manning, C.D., Raghavan, P., & Schütze, H. (2008). Introduction to Information Retrieval.Cambridge University Press.

4. Salton, G., & McGill, M.J. (1983). Introduction to Modern Information Retrieval. McGraw-Hill.