Skip Navigation
Expand
Indexing keywords and phrases for answers
Answer ID 1914   |   Last Review Date 09/23/2021

How are keywords and phrases indexed for our published answers?

Environment:

Answers, Keyword and Phrase Indexing
Oracle B2C Service

Resolution:

When an answer is saved, the answer's content is indexed in the knowledge base, and the answer is instantly available for searching. Single words and 2- and 3-word phrases are eligible to be indexed for an answer. Two factors determine whether a word or phrase is indexed: words specifically excluded from searching and a limitation on the number of words and phrases that can be indexed for a specific answer. More about that below. 

URL and File Attachment type answers:  You can create two additional types of answers -- URL and attachment.

URL Answers - When you create a URL answer, an external URL address will display to customers on the customer portal. When customers click the URL, an external web page opens in a new window. URL answers are also indexed as searchable text on the customer portal. You can still define many of the same answer fields as in HTML answers. However, some fields (such as the Question and Answer fields) are unavailable when adding a URL answer. The exact fields available depend on the permissions set in your profile.

File Attachment Answers - You can add file attachment answers to link to information contained in another document. File attachment answers contain many of the same answer fields as in HTML answers. However, some fields (such as the Question and Answer fields) are unavailable when adding file attachment answers. The exact fields available depend on the permissions set in your profile. You can attach only one file for each answer. File attachment answers are also indexed as searchable text on the customer portal.
 

The sections below apply to standard, HTML answers saved within the Oracle B2C Service application. These answers include the Summary, Question, and Answer fields.

 

Excluded Words List 

exclude_answers.txt
This file contains common words such as prepositions, adverbs, and pronouns that are excluded from indexing and searching functions on the customer portal. If you have specific words that you do not want customers to be able to search on, you can add these stopwords to the exclude_answers.txt file to exclude them from being indexed. Stopwords must be uppercase and typed on separate lines.

Note: Words in this file are also excluded when using SmartAssistant to search for an answer.

To view the contents of the exclude file, use the steps below:

  1. Go to the Configuration items, select Site Configuration > File Manager.
  2. Select the appropriate interface from the Available Interfaces menu.
  3. From the Switch to: menu, select wordlist files.
  4. Click the file name to open the file and review the alphabetical list. The word list file that excludes words from being indexed is named exclude_answers.txt

Note: If you choose to remove words from the list, you will need to rebuild the phrases table for your answers. For more information on modifying the exclude file, refer to Answer ID 1309: Excluding specific words or phrases from searching.

 

Phrase Indexing Limitation

When dealing with answers that include lengthy content, there is a limit of 750 (500 for incidents) as to the number of unique words and phrases indexed for each answer. Words and phrases located towards the end of longer answers may not be indexed and as a result may not be searchable.

When adding and editing answer content, the application indexes words and phrases based on the fields they are used in. The content is indexed in the following order:

  1. Keywords
  2. Summary
  3. Question
  4. Answer
  5. Flagged file attachments

For long answers, you can enter important keywords and phrases into the Keywords field when editing the answer. This ensures that those words and phrases will be indexed for that answer and are searchable.

Note: When entering multiple words and phrases into the Keywords field, be sure to separate entries with a comma to ensure optimal performance. For more information on the Keywords field when editing answers, refer to Answer ID 1301: Keywords field when editing answers.

Another option when dealing with longer answers is to consider breaking the answer into smaller, separate answers so that more of the overall content is indexed across several answers. Answers with fewer words are not adversely affected by the phrase limitation and may be easier for an end user to find and comprehend.

Note that this phrase indexing limitation was implemented by design in order to manage the overall size of the phrases table in the database. In the interests of maintaining an effective and efficient knowledge base, we recommend that answers be composed as succinctly and concisely as possible. This is important not only for the benefit of end users who must absorb the information, but for increasing the accuracy of their searches.

An overabundance of indexed keywords often makes searches less accurate, not more accurate.