PFI provides consulting service to help solve issues using expertise in natural language processing and machine learning. Below are examples of what we can provide.
We can extract specific keywords (people names, company names, product names, etc.) from the data managed by our customers. Using such, we can create a custom keyword dictionary for a specific domain, and then fine-tune Sedue searching and recommendation features accordingly. In addition, other natural language processing methods can be assisted to solve customer issues or improve the user experience of the operating service.
Machine learning can be applied to emails, documents and other data to achieve automatic categorization of spam and topics/subjects. If you are using our search service, search results can be narrowed down by using category results as a document field.
Documents that are generated continuously over time (blogs, twitter, logs, etc.) may gather more attention (and use more data) at times. For example, when the World Cup takes place and is televised, blogs and posts relating to it will naturally increase dramatically. “Burst analysis” detects this kind of sudden increase in data by observing changes statistically.
We are also open to the idea of collaborating with customers to develop and research something from scratch.