Publication Date: 2016-Jul-19
The IP.com Prior Art Database
We describe a method for improving the relevance of search-engine results based on network analysis. The method filters search results by analysing the network of an organisation or professional network and prioritising results based on their closeness in the network to the searcher.
Page 01 of 2
Disclosed is a method for improving the relevance of search-engine results based on network analysis. Searching a document database or the web can often throw up the issue of too many results that fit the keyword criteria. The problem is that the first slice (page) of hits may not be a good match for what the user is looking for, and better hits may exist but be prioritized lower by the search engine. We propose a cognitive filter on search results that would improve the relevance of the search result weighting, based on network analysis.
The method works as follows (we provide two scenarios for illustration: scenario A that works using a company organization chart, and scenario B that uses an industry or specialist database):
1A. Prepare a graph that maps the logical organization chart for a company or organization, or:
1B. Prepare a graph that maps individuals who have specific qualifications or specialisms, based on a bounded set of data; for example for medical specialists from hospitals and medical institutions.
2. Store metadata for documents, including author name(s), editor names, citations from other documents.
3. User logs in and performs a search.
4. Calculate scores from searcher to authors of search results One way to calculate a network closeness score could be as follows: 4.1 Traverse the graph built in step 1 and assess the number of hops between the searcher and the primary author.
For example, score = 10 - no of...