Yes, the technology is currently used to find concept similarities in documents. Unlike other techniques, ai-one's approach can detect concepts within paragraphs. Moreover, it can find "ideas" within very sparse text. The solution is owned by our customers -- so the it would only disappear if all entities went out of business.
It is a general purpose technology -- much like a programming language or chipset. The use cases include any problem that requires machine learning of content. Current deployments of the technology include: mining social media feeds, forensics, legal compliance, electronic discovery (of legal documents and evidence), knowledge management, etc.
We typically do not tell what our customers are doing -- we license the technology so they can embed it within their solutions.
They say next to nothing about their underlying techniques. All they seem to be getting at is using "context" (which could easily be as simple as n-grams) instead of just word-bags - not exactly a new idea.
Our technology is proprietary. To our knowledge, we are among the first to offer an autonomic machine learning API. Unlike n-gram and word-bags, our technology detects the inherent semantic value of any string of byte-patterns (the computer science definition of words).
The new idea is that the system learns upon stimulation (like a neural net) while providing transparency and control to the developer.
7 comments
[ 2.0 ms ] story [ 29.7 ms ] threadI also imagine a number of commerce disciplines are going to like this contribution that can be used to enforce copyright...
And authority groups could also benefit.
I definitely would NOT recommend stopping the research, but I would be concerned about it suddenly disappearing from public oversight.
We typically do not tell what our customers are doing -- we license the technology so they can embed it within their solutions.
The new idea is that the system learns upon stimulation (like a neural net) while providing transparency and control to the developer.