Search Insights

[Feature intended for testing and development only]

[Works with Elasticsearch only]

Add the Search Insights Widget to the Search Page to inspect two things:

• The query string that’s constructed by the back-end search code when the User enters a keyword

• The response string returned from the search engine

Note: The JSON returned as a request string is pruned from several Elasticsearch query defaults for clarity. To see the full request JSON that Elasticsearch processed, adjust the Elasticsearch server’s logging.

In Liferay DXP 7.2, the Insights widget now adds the response string to the widget’s output, and the Enable Score Explanation option (enabled by default) prints a relevance score explanation for each returned result.

When a search query is processed, results are returned. The concept of Relevance determines how well results match the query. The Score explanation for returned search documents helps clarify seemingly odd results, letting you adjust the relevancy scoring process by making matches in certain fields count for more (boosting the fields is the term for this).

Inspecting The Search Query String

To see the Search Insights widget in action, navigate to a Search Page in your test server and add it from the Add menu ().

Figure 1: The Search Insights widget is helpful during testing and development.

Once you search for keywords that return Search Results, the Search Insights portlet displays the returned query string in all its glory.

Figure 2: The full query string isn't for the faint of heart. This example is clipped to spare the reader.

Explaining Search Results

To enable or disable the Explain option,

1. Open the Search Insight widget’s Configuration screen.

2. There’s just one option: Enable Score Explanation. It’s a boolean field that’s enabled by default. De-select it to disable the explanation of each result’s relevance score.

Under the hood, the Explain option in the Search Insights widget is exposing an Elasticsearch API: Explain. See the Elasticsearch documentation for more details.

Here’s an abbreviated portion of the scoring explanation for the Search Document of the Test Test User when the searched keyword was test:

```json
_explanation":{  
   "value":9.461341,
   "description":"sum of:",
   "details":[  
      {  
         "value":9.461341,
         "description":"sum of:",
         "details":[  
            {  
               "value":1.0,
               "description":"emailAddress:*test*",
               "details":[  

               ]
            },
            {  
               "value":5.0,
               "description":"userName:*test*^5.0",
               "details":[  

               ]
            },
            {  
               "value":0.72928625,
               "description":"sum of:",
               "details":[  
                  ... 

            { 
               "value":1.0027686,
               "description":"sum of:",
               "details":[  
                  ...
                  {  
            {  
               "value":0.72928625,
               "description":"sum of:",
               "details":[  
                  ...
            {  
               "value":1.0,
               "description":"screenName:*test*",
               "details":[  

               ]
            }
         ]
      },
      ...
   ]
}}]}
```

Now you’re able to see the entire query string, the response string, and how each returned Search Document was scored.