Natural Language
and Content Processing

The central role played by language is also clearly reflected in industrial organisations, scientific and technological processes, public institutions as well as commercial enterprises.

Language is the crossroad of knowledge, communication and learning as all the information human beings are exposed to synthesize as well as exchange appears sooner or later in linguistic form. The assets of any of these working units is in fact strictly related to the body of documents they use to establish their organisational settings (e.g. internal policy or rule books), share the internal technical know-how (manuals, metadata schemas, project reports), design and impose internal process architectures (such as in business process modeling systems, tools and repositories) and finally, but not less importantly, to promote the outcomes outside the organization (through publicly available documents and marketing communication actions). All the above assets embody the core richness and specificity of an organization at a strictly linguistic level.

NLP is a crucial tool for content interpretation and knowledge acquisition against a large set of modern sources of information

Reveal promotes the empowering of all these organization-specific assets through a flexible and highly accurate language processing technology able to support machine reading (information extraction, automatic metadata creation, indexing), knowledge acquisition (knowledge discovery, fact extraction, semantic interpretation and predictive reasoning) as well as human-machine interaction as in automatic dialogue over the company texts, within the company customer community as well as towards the entire Web.

Modern NLP systems are able to adapt themselves through unsupervised Machine Learning methods

Researcher and scientist responsible of the foundation and development of Reveal have contributed to the study of robust and adaptive technologies for Natural Language Processing (NLP) that make this latter viable for a variety of applications ranging from semantic search to sentiment analysis, from advanced text analytics to real time marketing, NLP is a crucial tool for Content interpretation and knowledge acquisition against a large set of modern sources of information and knowledge: open data sets, standard Web contents, Social Web information as well as user-generated data.

By the exposure to such huge amounts of data, modern NLP systems are able to adapt themselves through unsupervised Machine Learning methods, to distill the domain knowledge in several effective ways and to use such knowledge to support a large number of semantic inferences and organization-specific decision-making processes. Notice that semantic inferences about linguistic data are widespread in modern applications such as ranking in search and recommending systems, market analysis through sentiment analysis over user-generated contents, compliance verification to norms, technology monitoring for competitive analytics, business process design as well as conversational software.

As the figure below suggests NLP allows to create value out of linguistic data through the effective chain:

The above chain distills value from an organisation’s documental assets, support the rationalization of documents and knowledge through the support of effective and justified processes and finally the analytics functions that allow optimizing these latter.

Atypical NLP Chain to Anticipate Users Needs

A typical NLP chain allows to recognize and interpret the facts and participating entities that a text describes, usually rooted in the real world to which the text makes reference to. In this way NLP allows us to understand contents as well as to anticipate user needs and personalize most of the organizational decisions (e.g. in advanced CRM).

Several Applications and Domains

The NLP engines engineered by Reveal have been successfully used in several applications and domains.

Flexibly released as Java-based language processing services mainly for:

• Robust morphological and grammatical analysis of texts written in different EU languages (English, Italian, Spanish, …)
• Information Extraction for the recognition and extraction of fine-grain phenomena such as mentions to events, people and organization or relational information about them
• Automatic metadata creation from texts such as editorial classification, organisational and compliance analysis, sentiment recognition or recommending
• Automatic domain and organization specific ranking functionalities in response to search or recommending requests
• Automatic Ontology population from texts
• Language specific semantic alignment between independent ontological schema for advanced and cost-effective data integration
• Question Answering and Interactive question-answering Services for dialogue-driven search
• Conversational Agents frameworks for automatic response machines as front ends for complex Web services (e.g. banking utilities, infomobility, voice-mediated interfaces)