Natural Language Processing
One of the most promising "new" New Technologies is Natural Language Processing. This discipline is on the border between computer science and linguistics, and applies to all aspects of human language. Several NLP applications have led to commercial software, widely disseminated:
- Machine translation, since the 1950s,
- Spell checkers,
- Information retrieval (search engines),
- Speech recognition (standard feature in the latest Windows releases),
- Speech synthesis of (GPS voice),
- Automatic text generation (letters sent to customers, weather),
- Text summarization…
In addition to these "vintage" applications, new categories of NLP software now take their place in business, in many different areas, with a significant Return On Investment:
- Automatic routing of inbound emails,
- Candidate screening and filtering on job boards,
- Information extraction,
- Text mining…
These technologies can also be applied for vertical solutions. For example, lawyers can automatically receive judgments classified by category, etc.
Current products that include a semantic approach (typically search engines) are often based on statistical methods. This approach provides good results, using well known Machine Learning techniques. However, it does not provide any real understanding of document content. In machine translation, statistical systems do not try to understand the original text. Ultimately, these applications process the text with poor language skills.
Semantic analysis and text understanding still represent a real technological pitfall that we want to solve. We have developed a "second generation" semantic analysis engine, combining linguistic rules with the statistical approach.
Semantic Web
The Semantic Web represents one of the most exciting evolutions of the Web. It aims to make web data visible, not only to huma eyes, but also to computers. To give an analogy, think of the barcode on a cereal box: a human will read the text written in the box, and the computer will read the content of the barcode. The Semantic Web also aims to provide reasoning capabilities to applications.