Search code examples
azureazure-cognitive-searchhit-highlighting

Azure search hit-highlighting and match delimiter


I am using hit-highlighting in azure search. It works fine but I want to fine tune it a bit.

Say, a field has the following value: "It uses period as the delimiter. If not, please clarify"

If I search for "please" I will get a highlight hit on that field, e.g.: "If not, <em>please</em> clarify"

If I search for "period" I will get a highlight hit on that field, e.g.: "It uses <em>period</em> as the delimiter."

After trying it with several examples it seems that it uses period (".") as a delimiter so that it doesn't return the whole field.

From another SO question (Hit Highlighting in Azure Search Service) it seems that I cannot configure azure search to return the whole field with all terms highlighted.

I want to ask:

  1. if this is really the case or more complex rules apply
  2. do I have any control of how the field is split for hit highlighting, e.g. change the delimiter to say "," or "\n"

Thanks in advance


Solution

  • Unfortunately there is no way to customize how documents are split for hit highlighting. Feel free to use Azure Search User Voice website to post improvements ideas giving other users opportunity to vote for them and helping us prioritize: http://feedback.azure.com/forums/263029-azure-search

    The hit highlighter splits documents into sentences. In general it's fair to assume it breaks on dots but it also handles abbreviations etc.