Why is it in the Keyword Search report, the words and phrases are frequently misspelled, i.e. "interfac" or "upgrad", and sometimes I see "<invalid query>" ?

Environment:

Phrase Searching, Analytics
Oracle B2C Service, All versions

Resolution:

The Keyword Search report does not show keywords and phrases exactly as your customers typed them. The entries listed in the Keyword Search report are the stemmed versions of the words and phrases entered into the Search field on your end-user pages.

Stemming allows the database to search on different variations of a specific word. For example, if you search on the word "typing", the database is also searched for type, typed, types, and retype, and answers that include any of those words are returned in the search results. In this case, the stemmed root of this word is "typ" which is what is listed if you generate the Keyword Search report.

Typically, if the word in the report is misspelled at the end of the word, such as a missing "e", "s", or "ing", or when the stem ends in an "i" instead of a "y", this is due to the stemming algorithm. However, in cases where the word is misspelled in the middle of the word, such as "signitur" or "priveledg", this indicates that the word was misspelled during the search.

In the Keyword Search report, it is not possible to get a list of non-stemmed words and phrases that were searched on. Only the search string that is used by Oracle B2C Service to search the knowledge base is included in the report.

If a search is not performed on a search string, either because the words in the search string are in the stopwords list or the string is composed entirely of punctuation, then the Phrase Stem listed will be "<invalid_query>."

For example, if the words "technical" and "support" are listed in your answers stopwords list, and a customer searches for "technical support," then the Keyword Search report will show a Phrase Stem of "<invalid_query>" not a Search String of "technical support."