User:Kingsif/interwikilinking tool ideas

Ideas for creation of a tool like suggestbot, which suggests parts of an article where a wikilink could be added, increasing navigation etc, based on finding articles through the "what links here" tool (which this page refers to as "original article"); finding the sentence which links (sentence only to prevent picking up unrelated data); and searching for similar terms in the original article like earwig's copyvio tool.

Based on sample from an article I have been working on recently, Dianna Agron (in table, "Article A"), and the current "what links here" tool.

The tool could be asked to look for shared wikilinks before similar terms, though this may have limited use, as wikilinks are not always present where there are currently reciprocal links, and many suggestions based on sharing other wikilinks are poor.

It could also simply look for the (target) article title, like the orphan tool. But, before looking, it should first check to see if the (target) article is already linked, which seems obvious (and which I think that tool does), but is worth stating. This would be both to prevent adding duplicate links and to prevent restatement; in the example table (row all in bold), the article Burlingame, California has a reciprocal link with Dianna Agron, but there are no similar terms in the text around the links at Article A, while there are similar terms in a different sentence that does appear related - this could lead to addition of a link to the article there, causing restatement. Looking for the article title as text can also create anomalies; dates and years will be identified but should not be linked, while (seen with Bob cut, row all in bold) not all links or potential links will use text identical to the article title. The orphan tool searches references, too, the DYK tool searches only prose text, it may be useful for a link finder tool to not search references, but look in all other text**.

User discretion should also be warned, as there will be list articles and other articles that do not warrant reciprocal links. The tool could discount list-class articles, but this would not stop all bad suggestions. Equally, the tool will not pick up all potentially-good suggestions: as seen with the row for Alexander McQueen (all in bold), the sentence at the article does not contain much context and there are no similar terms to be found in Article A, so no location text would be suggested, but there is a potentially-appropriate other location in Article A (both discussing the play McQueen, but in this instance the related data is in nearby sentences).

The tool could also exclude searching for the title of Article A, as it is unlikely to be helpful.

Reciprocal link present = green, Not present = red, Target article is list-class = yellow