The semantic analysis module is an optional module which uses semantic enrichment and natural language processing to add structure to unstructured data so that entities, terms, and relationships may be identified.
There are hundreds of contextual outcomes that can be discovered by text mining, like detecting objection in emails exchanged between a sales executive and a prospect or identifying a competitor that a sales executive had missed through document analysis. Another use case where semantics provide critical context is the analysis of tonality in customer support or sales interactions.
By analyzing press releases, the semantic engine is able to search for facts or relationships. An example of how this can be effective is when the semantic engine was able to identify that a given company had raised funds when a previous sales opportunity was stalling due to lack of resources. The application then suggested that the team use this fundraising concept to engage a current prospect in the same situation.
bee4sense solutions include advanced semantic, indexing and search functions, providing users with text mining capabilities.
These semantic analysis technologies work on three levels:
Word level (term analysis): the system analyzes the relevance of specific terms in the document corpus using a keyword-based statistical analysis method.
Sentence level (morpho-syntactic analysis): the system analyzes sentences for different types of words (noun, verb, adjective, complement, etc.) to help identify information or entities.
Discourse analysis: the system treats the text like a graph and analyzes relationships between entities cited in distant paragraphs, even in very long documents.
The bee4sense semantic indexing engine is capable of conducting all three types of analysis.
Finally, the integration of the semantic engine into SenseBuilder also means that it is possible to make corrections and apply these corrections to both the semantic rules and the indexed history.
The following semantic functions are available:
– morph-syntactic analysis, to identify a particular type of word (verb, noun, adjective) and its lemma (the standardized, gender-free, number-free, and inflection-free form of the word)
– recherche de termes en fonction de leur position absolue ou respective ou en termes de navigation dans un document
– term searches based on the absolute or respective position of the term found by browsing a document
– named entity extraction (organizations, individuals, job titles, locations, currencies, dates, etc.)
– terminological analysis, or extracting terms and creating glossaries to identify relevance
– terms in a particular field and structuring these terms based on simple relationships (for example, computer => desktop computer, laptop computer)
– concept relationship analysis (parent/child relationships, subsidiaries, competition, etc.)
– information or event extraction (calendar, politics, economics, etc.)
– sentiment analysis and opinion mining
– categorization using a module based on a pre-defined classification plan (supervised classification)
– document clustering (unsupervised classification)
– trend analysis (statistical frequency of a concept or term over a given period of time)