Home  Products Web Search Introduction
WEB SEARCH: INTRODUCTION
Relona is a search-engine technology startup headquartered in Nashua NH, USA with R&D in Bangalore, India. Relona has developed Intent-Analysis algorithms that can improve the accuracy of Web Search. Comparison studies indicate that when an existing engine such as Yahoo! Search is enhanced by Relona’s technology, it becomes as accurate as Google.
Intent-Analysis
Relona’s algorithms analyze the user’s search query to discover the search-intent. The data from this analysis is stored in an Intent-Map. The Intent-Map is a data-structure that specifies how the different keywords in the query should be selectively emphasized and selectively de-emphasized for best results. In addition, the Intent-Map also specifies how words should be grouped in multi-word concepts.
Query Modifications
Relona's algorithm is currently implemented over Yahoo's search API. Though the best way to use the intent-map is to integrate it directly into a search-engine, Yahoo's publicly available API does not support such integration. Therefore Relona's system is implemented as a query modifier. Relona's algorithm first analyzes the user's search query and generates an intent-map. Next it uses the intent-map to modify the user's search query. When this modified search query is run on Yahoo's engine, it produces superior results. The results of comparison studies indicate that query modification can help Yahoo overtake Google in terms of relevance.
Examples
The first example is about a search user who wants to find the director of the movie called Business is Business and enters the search query:
business is business director
Since Yahoo's search-engine does not explicitly identify the multi-word concept business is business, the user finds the results irrelevant. But Relona is able to improve the relevance of the results by modifying the query as illustrated in the side-by-side screenshots below: (Click here to zoom the screenshots)
 |
|
 |
The search user wants to find the director of the movie called Business is Business and enters the search query: business is business director
Since Yahoo's search-engine does not explicitly identify the multi-word concept business is business, the user finds the results irrelevant as shown above. |
|
|
| When Relona's algorithm analyzes the same query, it recognizes the concept Business is Business and applies appropriate quotes. When Yahoo's search-engine is run on the modified query "business is business" director, the results are relevant as shown above. |
|
The next example is of a search user who wants to find the scores and game statistics of the Australian basketball team called South Dragons. As illustrated in the side-by-side screenshots below, Relona is able to help Yahoo improve the quality of its search results: (Click here to zoom the screenshots)
 |
|
 |
In this example, the user wants to find the scores and statistics of the Australian basketball team called South Dragons. Unfortunately Yahoo's results for the query
south dragons score statistics are not very relevant as shown above. |
|
|
| Relona's intent-analysis algorithm recognizes South Dragons as a concept and adds quotes. Yahoo's results for the modified query "south dragons" score statistics are relevant as shown above. |
|
FURTHER READING
Demonstrations: A partial demo of Relona's technology is available online. Contact us to get the full demo.
Comparison Studies: Blinded 3-way comparison of Yahoo, Google, and Relona indicates that when Yahoo is enhanced with Relona's algorithm its search results are more accurate than Google.
FAQ: Answers to common questions about Relona Web Search.
|
WEB SEARCH 