Query Expansion

Voyager's Query Expansion can help you find additional relevant documents by including search terms that are similar to the keyword(s) you enter.  It finds these similar terms using Wikipedia.com as a reference. This word reference collection can be fine-tuned or even replaced with one that more closely matches your data.

For example, if you search for roads, you might find documents related to freeways and highways. If you search for a two-word term like climate change, you might get back terms like greenhouse gases,temperature increase or sustainability.

How it Works

To enhance your query, Voyager takes the search term(s) you enter and calculates other, similar terms.  It then uses these expanded terms as well as your original search term when it executes a query.  You can adjust how similar these extra terms must be on the Query Expansion management page (see below).  You can also remove any terms that may not be helpful to you.

 Keep in mind that this the calculation of similar terms can be highly sensitive to the similarity threshold you specify, so some experimentation is usually needed.  

Note:

The Query Expansion functionality only applies to text entered in the What are you looking for? box. It has no effect on terms in the Where is it located? box.

Using Query Expansion

Query expansion is disabled by default.  

  1. To enable it, click the toggle in the search box, either on the Home Page or Search Results page

  2. Enter your search keyword(s) in the What are you looking for? box and press Return or Click the Search icon. For example, if you enter water and click Search, you might see the additional terms shown below

  3. You can remove any of these terms by clicking the X next to it before you run the query again.

 

 

 Query Expansion Settings

You can specify a different engine for calculating similarity and you can also adjust the similarity threshold for query terms:

  1. Go to Manage Voyager > Index > Query Settings > Query Expansion

  2. Adjust the Similarity threshold (minimum 0; maximum 100) by entering a new value in Set Minimum Similarity Score (30-50% are recommended).  Any terms that fall below the threshold will not be used in the expanded query

  3. You can enter a new similarity engine by clicking Edit next to Wikipedia Query-Expansion Model. Usually this will be one based on your particular data set and data types.

  4. Click Save when you are done