Artificial intelligence engine STILINGUE – Blip | Blip Help

Index:

Introduction

Understand what the STILINGUE artificial intelligence engine is and how it works, along with its evolution in Natural Language Processing.

Always striving to facilitate access to millions of Portuguese-language information and conversations, STILINGUE has been using Artificial Intelligence (AI) to deliver real-time accurate data. To do so, it utilizes STILINGUE's proprietary Natural Language Processing (SNLP).

Natural Language Processing (NLP), is an Artificial Intelligence (AI) engine whose goal is to build a framework capable of understanding the characteristics of our written language and converting them into a computational language. This computational language is then interpreted by a machine/computer. Thus, the information received is automatically transformed into structured data (tables, numbers, and charts, for example), such as public opinions data on products, services, brands, politics, and more.

The Natural Language Processing (NLP) by STILINGUE, or SNLP, is the exclusive version created by the STILINGUE team of experts to perform this processing - translating the Portuguese language into a computational version through language processing. Within the platform, it can be found in the automatic sentiment classification, the term relevance system, automatic gender detection, automatic language filter, and the AAA Influencer Ranking.

Observe in the illustration below how the system processes the collected words: through NLP, words can be grouped, filtered (removing repetitions like prepositions), and interpreted. With the STILINGUE's artificial intelligence engine (SNLP), in addition to high-accuracy data classifications, the chart also shows term correlations, zoom capability, and other functionalities.

Our artificial intelligence engine brings significant improvements to how the STILINGUE platform performs this processing:

Reading Words and Expressions

The Portuguese language has many peculiarities and is highly dependent on context to establish meaning. Our artificial intelligence engine brings much more precision to reading, taking into account specific language expressions and its sentiment accuracy. For example:

“Preço salgado”; “Vale a pena”

The STILINGUE engine provides greater accuracy in grouping words to construct expressions, returning more informative terms, as the AI algorithm considers the context of each word.

Without word grouping: [final] [day] ...

With our artificial intelligence engine and word clustering: [end of the day]

Emoji Processing

Emojis also communicate a lot! As used in countless internet posts, STILINGUE saw it as an opportunity to evolve our artificial intelligence engine.

Note: in textless posts - where the content is composed solely of links, emojis, user mentions, numbers, email, or insufficient amount of letters - the language classification will be carried out through information from the language available in links, mentions, or hashtags.

In the absence of such information, the language will be classified through metadata, i.e., data that has some relation to the main post, such as the location and biography of users on a social network.

Polarity by Grammatical Category

As an essential part of the functionality evolution, our artificial intelligence engine brings higher accuracy, considering punctuation, grammatical class, word positioning, and much more.

"The week was busy, so we couldn't find the time to watch the night race in the urban center."

Therefore, the classification of publications is much more accurate, with a considerable reduction in Neutral and Unclassified posts, and bringing an expanded view of correlated terms. Currently, the accuracy for sentiment classification in PT-BR publications ranges from 75% to 82%. Learn more details about this metric in the FAQ.

For more information, visit the discussion on the subject at our community or videos on our channel. 😃

Related articles