It is not everyday that you come across search technology that challenges the established leaders in the search space – that too on the grounds of relevancy.
Relona, a start-up aims to do just that. With demonstrable statistics, the firm has proven that there is indeed room for radical technological improvements from newcomers in the search engine game.
Document Space Vs Query Space
It is a fact that there is an explosion of content on the web. Relona’s technology is based on the theory that the keyword space (or query space) has to grow in order to cover the whole swath of content on the web. This means that on an average, users will have to enter more keywords to get more relevant results from the search engine.
On the major search engines, the results returned for long queries is not as relevant as those returned for short queries. This is where Relona’s Intent Based Search algorithm uses statistical models to better map content to search queries by adjusting the emphasis on the words that form the query.
Intent Based Algorithm
The Intent based algorithm does computationally what the user does by habit when the results returned do not meet the relevance criteria – use different words to convey the same meaning. This is where Relona’s technology takes the middle path between natural language processing and using pure link analysis. By using statistical models, Relona’s search technology “guesses” users intent by adjusting the weight on different terms used in the search query.
Sprucing up relevance as compared to Google
What is really interesting is the enhanced relevance that MSN, Yahoo and Ask.com have (as compared to Google) when combined together with Relona’s Intent based algorithm.
- Ask.com – Relevance improves by 25% for two or more keywords.
- MSN-Live – Relevance improves by 20% for two or more keywords.
- Yahoo – Relevance improves by 35% for two or more keywords, making it 5% more relevant than Google.
The results are provided at Relona.net as impressive presentations and were obtained from analysis of AOL’s query logs. Also to be noted is that the above analysis was done without any direct integration with the search engines (Is this a hint they are open for acquisition?)
Relona’s technology is all the more alluring because it seeks to build upon the base of the first era of search engines and uses a statistical model that improves with usage. A demonstration of the search engine’s technology can be accessed from here.