Intellexer Sentiment Analyzer is a powerful and efficient solution that automatically extracts sentiments (positivity/negativity), opinion objects and emotions (liking, anger, disgust, etc.) from unstructured text information. Besides, Intellexer Sentiment Analyzer can be successfully used for document sentiment classification and review rating prediction tasks.

At present due to the intense increase of opinionated texts like product reviews, tweets and blog posts, the development of systems that provide solution for a wide range of sentiment analysis problems becomes more essential.

There are two main approaches to sentiment analysis: the lexicon-based and the learning approach.

The lexicon-based approach calculates the semantic orientation of words in a text by obtaining word polarities from a lexicon. The supervised learning approach uses statistical machine learning techniques to establish a model from a large corpus of documents. A set of sample opinions forms the training data from which the model is built. Applying machine learning techniques can achieve over 80% in sentiment accuracy. Unfortunately, this level of accuracy is not quite on par with the performance for full opinion text understanding.

The main factor that makes sentiment analysis difficult is that the structure of opinions is often context-sensitive and domain-dependent.

The team of R&D department of EffectiveSoft Ltd. has developed a unique hybrid approach to sentiment analysis which, unlike existing methods, is based on using not only linguistic and statistical information, but also a set of complex semantic rules. This technique provides one of the best sentiment accuracy and the system adaptation to the knowledge domain results on the market.

Our approach to sentiment analysis successfully process:

• Context-dependent opinions (short battery life → short recharge time)
• Domain-dependent opinions (small size (memory in Gadgets) → small size (hotel))
• Negatives and Intensifiers (not a bad service, too cheap device)
• Connectors (Although this hotel has brilliant location, it has a horrible service - negative opinion)
• Non-opinion phrases containing opinion words (romantic prose, stained glass)

Our experts create custom built ontologies for the most significant and widely used data domains (hotels, restaurants, gadgets, etc.). In order to obtain informative reports opinion, objects can be grouped in different ways according to the client’s preferences.