Reputation: 21
I'm asking about conceptual search (or concept-based search) which doesn't find documents based on specific word or phrase, but it returns relevant documents which are conceptually related regardless of shared terms or even a common language.
example: if you use the keyword "BMW" the results would contain "expensive cars" and so on.
I already have search engine based on Apache solr. now how to implement the conceptual search and if there are dictionary of concepts to use in my search engine so when can i find it ?
Upvotes: 1
Views: 903
Reputation: 5706
For that kind of intelligent search you should take a look at machine learning (to complement your search engine functionality), cuz' you need to make sense of you corpus and, when queried, suggest terms similar by meaning/semantic (not by Levenshtein distance).
Solr and other search engines are not meant to understand the data, they simply allow you to do advanced analysis and transformation of it. AFAIK the furthest you can go with search engines is to use synonyms, but that wont be enough for your case, right?
Take a look at Apache Mahout, Apache UIMA and Apache OpenNLP.
Upvotes: 1