4.5 Different types of text queries in Texis

There are five different text-retrieval methods/clauses within Texis.

  1. LIKEP The most common clause. It provides the most detailed and accurate ranking of the output and as such requires the most amount of CPU time. LIKEP factors the following match attributes when calculating the value provided in the variable.
    Set frequency in corpus
    Set frequency in document
    Set match cluster proximity
    Set match cluster ordering
    Set falloff distance
    Match cluster proximity to begin of field

    The relative importance of each of these factors may be adjusted prior to any query by using the SQL "set likep*" commands. Here are the list of LIKEP's controlling factors: likepallmatch likepdocfreq likepindexthresh likepinfthresh likepleadbias likepmode likeporder likepproximity likeprows likepstartwt likeptblfreq likepthresh likeptime

  2. LIKE The other commonly used text retrieval clause. It does not sort the output by rank, and only generates an estimate of rank to look at. It either returns the record, or it does not. It is used when you want to find all records matching a query and don't care about rank ordering.
  3. LIKER this is the first stage to LIKEP. It does not do any proximity ranking, relying instead on the absence or presence of the terms, and their relative weights. While faster than LIKEP it is not particularly useful for short queries, although it may be useful for large queries with lots of terms, when you are less concerned with proximity, word ordering and lead bias, and only want to know how many occurrences there are. likermaxrows likermaxthresh likerpercent likerrows likerthresh likerwt
  4. LIKE3 performs the index portion of a LIKE query only. Guaranteed to never post-process, although it does still process the new record list. Will perform as well as or better than LIKE. LIKE will be equivalent to LIKE3 if the index can resolve the entire query, otherwise LIKE3 may return matches that LIKE wouldn't.
  5. LIKEIN is an inverse of LIKE, used for profiling and routing, which we will discuss in depth later.
Back: Query processing options Next: Index types
Copyright © 2024 Thunderstone Software LLC. All rights reserved.