Search Features — Relevancy-Ranking


USA Daily Chronicles: Text Search and Relevancy    

USA Daily Chronicles: Text Search and Relevancy


dtSearch can sort and instantly re-sort search results by:

  • number of hits, file name, file date
  • metadata like subject or email recipient or sender

Natural language algorithms provide automatic term weighting, following a "plain English" or unstructured indexed search request.

  • Vector-space term weighting sorts by relevancy using the frequency and density of hits in your files.
  • For example, in the search request get me Sam’s memo on the 1999 CorpX takeover, if 1999 appeared in 3,000 files, and Sam appeared in only two files, then Sam would get a much higher relevancy rating, taking you straight to the most "relevant" files.

positional scoring option works with dtSearch's natural language relevancy ranking to rank documents more highly when hits are near the top of a file, or otherwise clustered in a file.

Variable term weighting can also apply to natural language and other queries.

  • Positive term weighting can place extra emphasis on one or more words: soup:8 or recipe:3
  • Negative term weighting can assign negative emphasis to one or more words: red or green or yellow:-7
  • Variable term weighting can also apply to metadata:  (description:5 contains (apple and pear)) or (author:2 contains smith)

See also Search Reports in the Forensics and Other Advanced Search topic

Developers, for more API-driven sorting and relevancy-ranking options, see Faceted Search and Data Classification