Data Ecosystem
for your Organization

The Sentiment and Trend Revealer

The Sentiment Revealer (SentiRe) allows organizations to gather and extract subjective aspects related to the opinions and sentiment of users (writers and or readers) about a product, an idea or a service.

It acts on information sources that are publically available, such as blogs, micro-blogs and forums, in particular characterized by unstructured information, such as free-text messages, HTML pages or texts.

SentiRE allows creating personalized reports out of relatively free large-scale document collections, in support of the decision-making processes in business intelligence scenarios (e.g. banking or brand reputation on the Web). Such reports enhance the information extracted by enabling timeliness in predictive business intelligence processes.

The “hidden” feeling in the message is made explicit using dedicated neural classifiers

It is a distributed and scalable system, based on a service architecture capable of downloading texts and messages from the Web and Social Networks, processing them using advanced Machine Learning tools in order to express contents, feelings and emotions expressed in the written text. This semantic enrichment is automatically performed in combination with other engines from Reveal, in particular RevNLT and Revealer, enabling advanced Business Intelligence and Brand Reputation processes for an effective aggregation and retrieval of information.

A typical workflow of SentiRe is the automatic collection and analysis of messages from Social Networks (such as Twitter). Messages are downloaded, filtered according to topics of interest for the client and their “hidden” feeling is made explicit using dedicated neural classifiers. This is combined with automatic methods of time-based and community analyses in order to answer questions such as “How changed the idea of this community about this topic?”

Highly Modular and Scalable to reduce Cost

It is deployed as a Service Oriented Architecture that can be directly used by the final users through specific web-apps or by other applications by exposing standard web-services.

Like other engines from Reveal, SentiRE and all its components are entirely implemented in JAVA, including classifiers and language processing modules. It supports scalable data-lakes (such as MongoDB) and it can be easily installed in any environment with reduced costs. It is highly modular in order to efficiently provide a customized solution to the final user. In the Reveal service ecosystem, SentiRe typically embodies all services involved in the acquisition of data, semantic analysis of collected messages and documents and community-based aggregation.
All these processes as summarized in the colored blocks from the following picture:

• Data interfaces: these enable the acquisition of material from the Web and the storage in dedicated data-lakes.
• Linguistic and Semantic services: these implement the linguistic analysis of input documents as well as the indexing of extracted contents and semantic information automatically derived by the Reveal’s processors.
• Data mining and semantic applications: the data mining services allow processing the entire message collections, together with the detection of communities generating such contents (through dedicated services of Community Analysis).
• Publication services: the publication services enable interaction with the final user, generally through Web-applications that can be customized according to the clients’ needs.