May 12, 2014 at 18:22Google Chrome extension to improve search quality
Almost 4 months ago, I made my first post on Habr, about how to make social likes useful for a programmer. Already a lot of water has flowed, the user base has stepped over 10,000 accounts, but the search, as one of the main functions of the application, remained unclaimed.
And there were several reasons for this - technical and non-technical. The technical problem was that the search (based on MongoDB full text search) was rather slow and did not give high accuracy (especially in Russian). But the non-technical component was more influenced, namely the UX problem.
What do we do when we want to find something? Right, open your favorite search engine. It’s difficult to break the habit and go searching for another site, even if you know that the results can have better characteristics. For a very long time this fact was hiding from us, but thanks to user feedback (thanks to sheershoff , dmytro_voloshyn , gimntut ) we decided to change the situation.
The first thing that was done was the migration of the search to ElasticSearch (which I will write a separate technical post about) and the second, the release of the extension to Google Chrome, which is able to expand the search results with data from Likeastore .
So Likeastore Chrome Extension was bornable to change the experience of using search.
As you know, the social component is gaining weight more and more in ranking algorithms for ranking results. Now, imagine that when you perform a normal search in Google, next to Google’s output, there is your personal selection of information, the relevance of which is extremely high, since when you already marked it as liked.
Actually, this is the main function of expansion to chrome. In test mode, he worked all last week and today he goes out.
You must have an account in Likeastore, install the extension in chrome and try to search for something.
We continue to work on the quality of the results. Also, we think about supporting other browsers (Firefox, Safari) as well as supporting other search engines (Yandex, Bing).
And there were several reasons for this - technical and non-technical. The technical problem was that the search (based on MongoDB full text search) was rather slow and did not give high accuracy (especially in Russian). But the non-technical component was more influenced, namely the UX problem.
What do we do when we want to find something? Right, open your favorite search engine. It’s difficult to break the habit and go searching for another site, even if you know that the results can have better characteristics. For a very long time this fact was hiding from us, but thanks to user feedback (thanks to sheershoff , dmytro_voloshyn , gimntut ) we decided to change the situation.
The first thing that was done was the migration of the search to ElasticSearch (which I will write a separate technical post about) and the second, the release of the extension to Google Chrome, which is able to expand the search results with data from Likeastore .
So Likeastore Chrome Extension was bornable to change the experience of using search.
As you know, the social component is gaining weight more and more in ranking algorithms for ranking results. Now, imagine that when you perform a normal search in Google, next to Google’s output, there is your personal selection of information, the relevance of which is extremely high, since when you already marked it as liked.
Actually, this is the main function of expansion to chrome. In test mode, he worked all last week and today he goes out.
You must have an account in Likeastore, install the extension in chrome and try to search for something.
We continue to work on the quality of the results. Also, we think about supporting other browsers (Firefox, Safari) as well as supporting other search engines (Yandex, Bing).