Context is key
Text understanding by SentiSquare

Problems of
textual communication processing

Problem 1
Languages are complex

  • Complex morphology
  • Polysemy
  • Slang expressions
  • Typos
  • Sarcasm and irony

Problem 2
Communication is distinctive

  • Product-specific slang
  • Specific problems customers are facing
  • Every company has its own language

Problem 3
Communication is dynamic

  • New trends
  • New topics
  • What was there 6 months ago is not there anymore
  • Unexpected problems

SentiSquare solution in 3 steps

  • General knowledge about the language is encoded in the pre-trained language model. This allow us to cover common understanding.
  • Unsupervised machine learning is used to discover common patterns in client data. General language models are tailored to understand the specifics of client data.
  • Supervised machine learning is involved to interpret the patterns in a desirable way. This allows us to deliver business-specific solutions for text-driven processes.


SentiSquare’s NLP technology is based on distributional semantics. This approach enables us to represent the meaning of a text without any supervision. The principle goes as follows: “You shall know a word by the company it keeps” (Firth, 1957).

Essentially, words are presumed to have similar meanings if they occur in similar contexts. That opens an opportunity for the quantification of meaning: textual expressions can be represented as vectors in a high-dimensional semantic space, encoding the distribution of words over contexts (this is how we got the term "distributional semantics").

Tailor-made models

State-of-the-art semi-supervised machine learning techniques allow the training of custom NLP models directly on company-specific data, reaching human-like accuracy at highly optimized computational costs. Language-independent contextual patterns also make our technology suitable also for underserved languages.

Distributional semantics

The secret ingredient of the SentiSquare No-Code NLP platform is distributional semantics. The main idea of distributional semantics is that "meaning follows from context". We will extract the right meanings from your texts. In minutes, your data will be sorted, categorised, and you'll know what they're holding.

Exception handling included

Built-in functionality for adding business rules to handle exceptions and combining models into a complex system allows companies to have text-driven processes under control during their daily operations.

Technological comparison
with our competitors

General (pre-built) models
Vocabulary-based approach
Bespoke NLP
No-Code NLP platform
Solution to my problem
  • Can the solution be tailored to my use case and my data?
  • Can the solution achieve human-like accuracy?
No special know-how
  • Can I use it even if I don't have a special IT or AI/NLP background?

Discover inspiring success
stories from our customers

Read about real-life experiences with the SentiSquare No-Code NLP solution.

Read more