👀 Views: 89 💬 Answers: 1 📅 Created: 2025-06-26

elasticsearch function-score scoring query json

I'm working through a tutorial and I've spent hours debugging this and I'm struggling with a function score query in Elasticsearch 8.5 where the custom scoring seems to be giving unexpected results... I want to boost documents based on a certain field, but the scoring does not appear to reflect the expected boost. Here's the query I've constructed: ```json { "query": { "function_score": { "query": { "match_all": {} }, "functions": [ { "filter": { "term": { "category": "electronics" } }, "weight": 2 }, { "filter": { "range": { "price": { "lt": 100 } } }, "weight": 1.5 } ], "score_mode": "sum", "boost_mode": "multiply" } } } ``` With this setup, I expected documents in the "electronics" category to have a significantly higher score compared to others, especially those priced under $100. However, when I run the query, the scores are not reflecting this boost accurately. For example, a document with a category of "furniture" is scoring higher than some electronics items, which should not be the case. I checked the mappings and confirmed that the `category` field is a keyword and `price` is a float, so I think the field types are correct. I've also tried using different score modes like `max` and `avg`, but the results remain inconsistent. The expected scores seem to be overridden somewhere. I even added a debug query to examine the breakdown of scoring, and I noticed that some documents with null or missing values in the `price` field are being scored unexpectedly high. Here's how I attempted to visualize the scoring breakdown: ```json { "query": { "function_score": { "query": { "match_all": {} }, "functions": [...], "score_mode": "sum", "boost_mode": "multiply" } }, "explain": true } ``` What might be causing these unexpected score calculations? Are there any best practices for configuring function score queries to prevent documents with missing values from skewing the results? I would appreciate any insights or examples from your experiences. Thanks for taking the time to read this!