Collective Project

HP Collective is an application for rapidly and optimally responding to enterprise opportunities and challenges, by leveraging the tacit knowledge in an enterprise, via identifying the right subject matter expert(s). HP Collective is a web application that automatically discovers experts and their expertise via linguistic and semantic analysis of their work products (e.g., e-mails, patents, papers, reports, presentations, and blogs). The system combines multiple data sources, including unstructured work products and structured sources such as Enterprise Directory and document meta-data, into one holistic model of organizational knowledge. The key feature of HP Collective is being "passive"; where the employees do not fill out or maintain forms or profiles.


  Demo

Try our our demo application of HP Collective.
Unfortunately our current demo contains HP confidential information, so this link is not accessible outside HP's network. We are working on a public demo. Stay tuned.

  Technology
Document Similarity Document similarity is computing using our own version of query expansion by pseudo relevance feedback, which has been shown to be effective for search engines. This version computes the expanding terms using a weighted correlation to the top ranked results, using their scores rather than their ranks only.
Graph Algorithms Graph algorithms identify sub-graphs of the entity-relation graph that contain various personalized views of the data. Thus, we can create view on-line that are relevant to query and personalized to the user. The graph algorithms allow flowing interest from a node to other nodes in the graph. Specifically, to find the relevance graph of a person, flow can be moved from the person, thru her organizational hierarchy, to the content they create and from there to similar content that is created in different places in the enterprise. This way, relevant people, content and topics can be explored.

  News & Publication