最好还是不结合搜索引擎和推荐系统? [英] Better or Not combine Search Engine and Recommend System?

问题描述

在我们的项目中，我们使用搜索引擎，但结果需要根据每个用户的兴趣进行排名，类似于根据用户的关键字进行推荐.

In our project, we use search engine, but the result need to be ranked based on each user's interest, similar to recommendation according to users' keyword.

如果我们将两个系统分开，会花费很多时间.

If we separate the two system, it would cost a lot time.

是否有更好的方法将搜索引擎和推荐系统结合在一起?

Is there a better way to combine Search Engine and Recommend System together?

或者是否有一种简单的方法可以自定义我的排名策略来实现这一目标?

Or is there a simple way to customize my ranking strategy to achieve this?

推荐答案

这也是我们在项目中尝试做的事情.解决这个问题有两件事 - 相关性与个性化.您应该查看有多少个性化会破坏查询的相关性.例如，如果我推荐新闻，那么根据位置进行推荐是有意义的.我希望您已经分析了用例.

This is what we were trying to do in our project as well. There are two things while solving this problem - Relevancy vs Personalization. You should look at how much of personalization is ruining the relevancy of the query. For example, if I'm suggesting news, then it makes sense to suggest based on location. I hope you already would have analyzed the use cases.

我遵循的方式是 - 在获得搜索结果后，重新排列结果以提供个人建议.例如，如果我正在搜索特定的算法来编码，然后获取结果集并根据我的偏好重新排名，可以说，Java(基于我以前的历史)将是有意义的.无论如何，相关性是最重要的，然后我们才能适应用户的喜好.

The way that I followed was - after getting the results on the search, then re-rank results to give personal suggestions. For example if I was searching for a specific algorithm to code, then getting the result set and re-ranking on my preference, lets say on, Java (based on my previous history) will make sense. In any case relevancy is of utmost importance and then we fit in user's preferences.

同样，用例很重要，如果这是用于新闻搜索，那么直接查询和检索位置是最好的方法.

Again the use case is important, if this was for a news search, then directly querying and retrieving on location is best way to do it.

这篇关于最好还是不结合搜索引擎和推荐系统?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！