User:Danilo Eidy/sandbox



KeyGraph is a method of processing data arrays and visualizing in a network graph. KeyGraph uses a genetic algorithm and processes building blocks to generate a solution to hard problems and create network graph. The method identifies clusters of associated elements of the data and hubbing elements that associate clusters, revealing stable patterns and rare elements with high connectivity in a network graph. Applied in various fields, KeyGraph summarizes groups of data and rare and impactful data for chance discovery. Some examples of data arrays are POS data, Bag-Of-Words, Relational Databases, network data, and others.

The Ohsawa's KeyGraph Method
The KeyGraph method consists of 2 major steps: 1) Clustering Highly frequent co-occurrent elements, and 2) Hubbing clusters with infrequent co-occurrent elements.

Step 1) Clustering Elements that frequently co-occur in arrays of data, will be extracted and pair wisely associated to each other. Elements with multiple associations will form clusters in a network graph, where the element is represented as a node and the association is represented as an edge. The associations between elements in clusters represent stable associations that repeatedly occur in arrays of data.

Step 2) Hubbing Elements that infrequently co-occur in arrays of data, will be extracted if they co-occur with at least one element of two different clusters. In the network graph, Hubbing elements are represented between clusters pair wisely associated with elements of two or more clusters with edges. Due to the connectivity among clusters, Hubbing elements are considered key nodes in the network graph. Hubbing elements represent chances in the interpretation of the data. Due to its rarity and connectivity, it suggests to bee an unexplored and potential element in the data.

KeyGraph and Language
In Computational Linguistics, the text is composed of arrays of linguistic data, where documents are decomposed into paragraphs, paragraphs into sentences, sentences into words and, as arrays, forms bags of words. When KeyGraph is applied to text, it generates semantic fields in clusters of textual data and key terms that reveal semantic relationships among semantic fields. For example, a collection of documents of various topics is used with KeyGraph. The method will gather highly frequent terms and generate clusters of terms in nodes, linked by edges. Unfrequent terms will link clusters when they occur with terms of the different clusters.

KeyGraph and Events
In Sciences, events are the objects of investigation of causal relationships of facts in the world. Sequences and co-occurrences of events are largely used to explain hypotheses in scientific terms. When KeyGraph is applied to events, it generates correlated events in clusters of facts and rare events that reveal latent causality among events. For example, let a collection of events be set in arrays in periods of time. The method will gather events with facts that frequently occur in the same period and generate clusters of co-occurrent events. Infrequent events will link clusters when they recur in different periods of time with events of different clusters.

KeyGraph and Behavior
In Psychology and Business, human behavior is the object of study to understand human as an individual, in society and how to make decisions. When KeyGraph is applied to behavioral data, it generates patterns of activities and unfrequent behaviors that reveal abnormalities or constrained decisions. For example, let supermarket's baskets in Point-Of-Sales (POS) data be arrays in each consumers' behavior. The method will associate products that are commonly purchased together and generate common baskets. Eventually, infrequently purchased products will show in common baskets, revealing an unusual behavior of the consumer but still associated with common baskets. The classic example of beer and diaper association would appear as a common association within a cluster in KeyGraph. In advantage, KeyGraph method shows the big picture and emergent associations of products, for example, the emerging consumption of new products or special products that trigger the addition of products from different common baskets.

KeyGraph and Other Networks
Networks can also be considered arrays of data, where nodes and links can be arranged in groups according to their connectivity, such as Hypertext networks, described as arrays of hyperlinks in a documents, Social networks as arrays of friends of each person, Electrical networks as arrays of electrical components, street network as arrays of street in a pathway, and so on. In Sociology and social psychology, collective behavior is defined in groups of people. Configuring collective behavior in social networks, KeyGraph reveals key persons such as influencers in weak tie theory. Also called bridges, this hubbing person connects closed social networks represented in clusters.

KeyGraph and Innovation
The versatility of applications of KeyGraph in different fields allows its applications that often require interdisciplinary approaches, such as creativity, innovation, political discussions, business strategies, and other hard problems. In creativity studies, KeyGraph supports thinking out of the box, stimulating individuals with visual information and constraints. Consolidated information in clusters helps individuals to retrieve their memory from their own knowledge, while hubbing terms bring awareness of relevant pieces of information that may not be directly retrieved but can be connected by active cognition and form new ideas. When applied to business, KeyGraph supports essential processes of innovation, such as creativity in product or service design   , and business and market strategies. Associated with Innovators Marketplace method, It stimulates individuals in the ideation and exploration processes of innovation.

Other applications
As a tool for discoveries in a complex context, considering multiple data, KeyGraph has being applied to risk prediction, such as earthquakes  , and terrorism , helping by increasing awareness on unpredictable, latent and hidden factors that may be relevant in risk prevention plans. Another application is the political speech analysis, where analyzer consider semantic relationships in the KeyGraph to interpret consistency, frequency, and dominance of ideas in political speeches and delineate social streams.