Semantic natural language processing interprets the meaning of free text and enables users to find, mine and organize very large archives quickly and effectively. Linguistic semantic processing finds all and only the desired information because it determines meaning in context and maps synonym and hyponym relationships. It avoids assigning incorrect relationships because meaning is precisely determined. At the same time it makes all the desired connections exhaustively because it is backed by a massive lexicon and semantic map. The key scalably of Cognition's linguistic semantic processing is bottom-up interpretation of the text, finding the meaning of words and phrases in the local context one at a time. The technology has a semantic map and algorithms that interpret language linguistically rather statistically, so that the meaning of a given document is independently determined. As a result the methods scale to a theoretically unlimited number of documents. Linguistic semantic NLP is being deployed in many applications that facilitate rapid and accurate management of very large archives. . 1. Free auto categorization - Texts are categorized into an existing ontology or a special client-defined ontology according to the salient concepts in them. 2. Segregation by genre - The software determines which of a predetermined set of genres a documents falls into. In the legal domain, the genre set might be "contracts" and within "contracts", "employment contract", "services contract", etc., PPM, "pricing proposal", "mortgage agreement", and so on.
3. Conceptual foldering (or tagging) - Documents are placed in conceptual folders using conceptual Boolean expressions that cover all of the topics desired for the folders. This is especially useful in e-Discovery, where documents can be culled leaving only the relevant portion to be reviewed.
4. Intelligent search - The semantic search function retrieves almost all and only the desired documents. Very high precision is achieved by disambiguating words in context, and by phrasal reasoning. Very high recall is achieved by paraphrase and ontological reasoning.
5. Text Analytics - Calculating the frequency and salience of words, word senses, concepts and phrases in a document or document base lays bare its significant semantic content.
6. Sentiment analysis - With semantic processing, sentiments can be determined. Existing lexical resources identify the "pejorative" and "negative" words.
7. Language monitoring - In some situations such as child chat or email, certain types of language may need to be blocked. Linguistic semantic processing detects undesirable (or desirable) language as defined by administrators.
3. Conceptual foldering (or tagging) - Documents are placed in conceptual folders using conceptual Boolean expressions that cover all of the topics desired for the folders. This is especially useful in e-Discovery, where documents can be culled leaving only the relevant portion to be reviewed.
4. Intelligent search - The semantic search function retrieves almost all and only the desired documents. Very high precision is achieved by disambiguating words in context, and by phrasal reasoning. Very high recall is achieved by paraphrase and ontological reasoning.
5. Text Analytics - Calculating the frequency and salience of words, word senses, concepts and phrases in a document or document base lays bare its significant semantic content.
6. Sentiment analysis - With semantic processing, sentiments can be determined. Existing lexical resources identify the "pejorative" and "negative" words.
7. Language monitoring - In some situations such as child chat or email, certain types of language may need to be blocked. Linguistic semantic processing detects undesirable (or desirable) language as defined by administrators.
Kathleen Dahlgren has a Ph.D. in Linguistics and a Post-Doc in Computer Science from UCLA. She has worked and contributed publications in computational linguistics for over 20 years. Her publications cover topics in sense disambiguation, question-answering, relevance, coherence and anaphora resolution. Her book, Naive Semantics for Natural Language Understanding primarily treats a method for representing commonsense knowledge and lexical knowledge, and how this can be used in sense disambiguation and discourse reasoning. The software offered at Cognition Technologies is patented by Kathleen Dahlgren and Edward P. Stabler, Jr., and has been under development for a number of years, so that it now has a wide coverage semantic map of English.
No comments:
Post a Comment