I recently joined the Intel Software Partner Program for a variety of reasons but one of these is to obtain input/information on market needs for a new product that we are considering.
For the lack of anything else I have named it a "Semantic Processing Unit".
Can I obtain feedback via this forum on the perceived value to the market?
It is a technical product as I envision it now so it was suggested that I post something in the community.
I provided a brief description at the bottom of this post.
Is this the appropriate forum, and could I get input from those who feel that they could use such a product? I would like to get a dialogue going, i.e., catch-ball to get into some details on who would buy it, what functionality is desired, etc.
I would be most appreciative of the input and discussion.
Semantic Processing Unit or SPU
Current Status: Prototype for medical information retrieval.
User Benefits: Very easy to use semantic interface and significant increase in search accuracy
100% semantic search (i.e., concept-based search) using ontologies and no keywords
Very high speed processing using complex natural language and topology-based algorithms that significantly improve search accuracy
Artificial intelligence reasoning to extract taxonomies/ontologies from free-form text; this can both enable and reduce the cost of semantic search engines/knowledge management tools
Applicable to any domain (prototyped in medicine due to readily available ontologies)
Highly parallel, many-core Larrabee architecture for very high-speed processing increased throughput and accuracy via highly-parallel processing
Note: this approach was recommended by an associate (PhD-level) who recently joined the Larrabee team
Likely Applications (more may be found during research):
Knowledge management products inside corporate firewall (server products, possible desktop in future)
Help desk/user help tools for technical topics, such as manuals and complex software applications (server-based with access via web, possibly desktop in future)
Internet search for fine-grained information extraction (server/grid-based products) specifics need further research