Back to results
US 8,751,218 B2Granted
Indexing content at semantic level
AssigneeSiemens Aktiengesellschaft
InventorsJiangbo Dang, Murat Kalender, Candemir Toklu, Kenneth Hampel
Life of this patent
Estimating the term…
Abstract
Systems and methods are disclosed that perform automated semantic tagging. Automated semantic tagging produces semantically linked tags for a given text content. Embodiments provide ontology mapping algorithms and concept weighting algorithms that create accurate semantic tags that can be used to improve enterprise content management, and search for better knowledge management and collaboration.
Claims23
A semantic tagging method that outputs semantically linked tags for text content comprising:
- inputting the text content to a computer;
- by the computer, extracting nouns and noun phrases from the text content;
- by the computer, mapping the extracted nouns and noun phrases to terms of an ontology;
- by the computer, mapping the extracted nouns and noun phrases to a correct sense of the ontology terms using lexical chaining Word Sense Disambiguation (WSD) algorithms;
- by the computer, weighting the extracted nouns' and noun phrases' concepts from their ontological and statistical features;
- by the computer, extracting key-phrases from the weighted concepts;
- and
- by the computer, outputting the key-phrases as semantic tags;
- wherein mapping the extracted nouns and noun phrases to terms of an ontology further comprises:
- accessing the ontology;
- comparing the nouns and noun phrases with the ontology terms;
- and
- if there is an exact match, retrieving the ontology term, and if there is not an exact match, performing phrase mapping using a rule based algorithm;
- and
- wherein the phrase mapping rule based algorithm further comprises:
- generating two sub phrases by removing the first and last word from the noun phrases and replacing the original noun phrases with the two sub phrases;
- and
- repeating the phrase mapping until there is a match or there are no more noun phrases remaining.
15 more elements
A semantic tagging method that outputs semantically linked tags for text content comprising:
- inputting the text content to a computer;
- by the computer, extracting nouns and noun phrases from the text content;
- by the computer, mapping the extracted nouns and noun phrases to terms of an ontology;
- by the computer, mapping the extracted nouns and noun phrases to a correct sense of the ontology terms using lexical chaining Word Sense Disambiguation (WSD) algorithms;
- by the computer, weighting the extracted nouns' and noun phrases' concepts from their ontological and statistical features;
- by the computer, extracting key-phrases from the weighted concepts;
- and
- by the computer, outputting the key-phrases as semantic tags;
- wherein mapping the extracted nouns and noun phrases to a correct sense of the ontology terms further comprises:
- creating a hypernym lexical chain with WSD scores for senses of the input nouns and noun phrases;
- creating a domain lexical chain with WSD scores for senses of the input nouns and noun phrases;
- creating a holonym lexical chain with WSD scores for senses of the input nouns and noun phrases;
- creating a category lexical chain with WSD scores for senses of the input nouns and noun phrases;
- normalizing the WSD scores in each of the hypernym, domain, category and holonym lexical chains;
- generating a WSD score using a sense rank algorithm;
- assigning weights to each WSD score and summing the weighted WSD scores for the hypernym, domain, category, holonym lexical chains, and the sense rank algorithm;
- evaluating the hypernym, domain, category and holonym lexical chains, the sense rank algorithm and optimizing their weights for a given target data;
- and
- mapping the nouns and noun phrases to a highest ranked sense.
16 more elements
Dependencies are inferred from the claim text.