Use Topic Maps for Searching

DEVONagent is a great product!

Would you consider adding a button in the browser to generate a topic map for a page and/or site, and then use the topic map to automatically build new searches (aka better queries)?

Once found, pages could also be filtered by dragging topics closer together or further apart. This would allow one to refine a search dynamically.

IMHO these two features would make DEVONagent a game changing product in the search world.

@ “pcafstockf” : This certainly sounds like a rather interesting and useful feature, one that I’ve never come across. Could you give us a site where we could explore a topic map? I’d like to find out more!



@ “twicks” : I think I have misunderstood either the DA interface, or your question.
I had assumed that the distance between nodes in DA’s “Topics Map” was determined by the averaged proximity of words to each other (some kind of suffix tree clustering).

By changing the “desired” distance between nodes, you could filter documents that had larger distances and in theory might be less relevant to your search.

My limited understanding of Topic Maps comes from:

And a limited understanding of STC from:

On a related note, some early search clustering work was done by “Grouper”

Please let me know if I missed the mark on your question.



Nope, no misunderstanding on your part. First, I misread “DA” as “DT” and I haven’t dug into DA very much.

Second, your links were quite helpful in providing further background.