Google’s Hummingbird update
If we use a search engine and get relevant results we are likely to use that search engine again and again. What causes people to switch search engines?
- The most influential cause of search engine switch is due to long tail queries (so 6 or more keywords).
- If a search engine processes a long tail query and returns low relevant results then this user is likely to switch search engine.
- The processing of long tail queries equally distinguishes search engines and good search engines are those that can return relevant results for long tail queries
Popular search engines have recently adapted to this. Google in September 2013 announced a new engine known as Hummingbird. The new engine does not make Google a semantic search engine but it does improve the processing of long tail queries. In light of Google Glass and mobile devices conversational search is going to be the next big thing for web search. Hummingbird is essential to improve long tail queries.
Google and Microsoft’s Bing are not semantic search engines, however, they are getting better at returning relevant results. If you were to carry out a complex search, for example [picture of blue flower with 6 petals with sea front background], on Google you will soon find that the results will not be highly relevant. Search engines are beginning to see value in improving the relevance of long tail queries because irrelavant long tail queries causes search engine switch, often permanently.
There are several wider reasons that directly influence search engine switch. They include: localised learning of the system; advanced search features; culture; previous encounters; presentation of results; precision and recall (see below).
Why search engines give lots of results?
Search engines can be evaluated using 3 metrics: precision; recall; and speed. Precision is all about returning relevant results. Recall examines the number of results and matches them to the actual query. Speed, thirdly, is all about getting lots of relevant documents that match our query in a quick manner. Search engines incorporate these metrics into every search. This explains why we get pages and pages of relevant results for our search queries (precision and recall).
Posted by Gerald Murphy
- Zaragoza, H. Cambazoglu, B.B. and Baeza-Yates, R. (2010) Web search solved? In Proceedings of the 19th ACM international conference on Information and knowledge management — CIKM ’10. pp. 529
- Zhou, K. Li.X. and Zha, H. (2012) Collaborative ranking: Improving the relevance for tail queries. CIKM ’12. pp. 1900–1904