In my last post I mentioned that similarity based recommendation engine in sifarish only considered categorical and integer attributes. I have added support for text attributes to sifarish . I am using Lucene for text processing and a variation of jaccard coefficient for distance between text attribute values. Lucene is an Apache open source package for text indexing and processing. In this post, I will go over the text analytic related enhancements. As a quick disclaimer: for this post, by text processing I mean the so called bag of words approach,
Read full article from Similarity Based Recommendation – Tossed up with Text Analytic | Mawazo
No comments:
Post a Comment