How to improve the accuracy of retrieval如何提高检索的准确率

[chaijl]

I uploaded a bunch of very standard JSON structured data, one of which is a date with a standard format of 2025-05-17 characters. However, when I searched for the data on May 17th, I found that the first few entries were all blocks containing 17 characters. Is there any way to improve this
我上传一堆很标准的json结构化数据，其中有个数据是日期的，标准的2025-05-17这样格式的字符，但我搜索5月17日的数据，查出来前几条都是那些里面含了17字符的块，这个有什么方法可以提升吗

[dosubot[bot]]

Hello @chaijl! I'm Dosu, a bot here to help you with bugs, answer questions, and support your contributions while you wait for a human maintainer. How can I assist you today?

要提高检索的准确率以便在搜索日期时能够准确匹配标准格式的JSON数据，可以使用以下的日期匹配配置：

{
  "dt": {
    "match_pattern": "regex",
    "match": "^.*(_dt|_time|_at)$",
    "mapping": {
      "type": "date",
      "format": "yyyy-MM-dd HH:mm:ss||yyyy-MM-dd||yyyy-MM-dd_HH:mm:ss",
      "store": true
    }
  }
}

这个配置使用正则表达式匹配字段名中包含_dt、_time或_at的字段，并将其映射为日期类型。支持的日期格式包括yyyy-MM-dd HH:mm:ss、yyyy-MM-dd和yyyy-MM-dd_HH:mm:ss。通过这种方式，可以确保在检索时准确匹配这些日期格式的数据^[1]。

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[ZhenhangTung]

@chaijl Hi, I’m the product manager for RAGFlow—thank you for raising this!

Improving retrieval accuracy is one of our core priorities.
To help us investigate and support you better, could you share:

• A few anonymized examples of your JSON data?
• Your current knowledge base configuration?

This context would really help us identify where the precision can be improved. Appreciate your input!

[chaijl]

The screenshot above is a configuration screenshot of my knowledge base. Everything else is not enabled by default, and the file contains JSON format data from one of my days
日报2025-01-01.json

[ZhenhangTung]

Really appreciate your detailed feedback. Just discussed this use case internally and I'll do a research and design it in the future coming versions.

[yingfeng]

Relevance has nothing to do with rules. What you wanted is to import Json into traditional databases and perform a SQL filter query. If you want a search(not query) together with rules, you can only specify those relevant fields such as date as dedicate fields within databases. However, RAGFlow does not encapsulate such scenarios because we don't know how to make it as a standard product design.

[ZhenhangTung]

We’ll temporarily close this thread for now, but feel free to reopen or submit a new one in English!

Thank you for your understanding!