Use text to make the right decisions

A lot of useful customer information is available (but hidden) not only in structured forms of relational databases and spreadsheets, but also in textual form in internal databases and external sources. The information emerging from a combination of structured and unstructured data sources represents a real gold treasure that can influence important business decisions in any organization. To discover patterns and trends in text data, it is no longer possible to classify them using manual analysis only. Ataccama introduces solutions for automatic information extraction from text data and combines it with structured data analysis, making the most of the rich experience and expertise we have within the data quality domain to improve your business processes.

Ataccama Text Analytics solution is a part of Big Data Engine, a powerful technology designed to tackle big data challenges. Ataccama Text Analytics contains all the functionalities important for a correct text preprocessing and text mining. While text preprocessing methods are applied to fix the data and make it suitable for further analysis, text mining extracts relevant information from unstructured texts.

Banking Case Study: Text Analytics Tenfolds Event Marketing Campaigns’ Success


  • Reliable fixing. Automated correction of grammatical and spelling errors text data usually contains.
  • Data parsing. Linguistic parsing of individual text data units using state-of-the-art tokenizer and rule-based sentence splitter.
  • Reduction and expansion. Having inflected or derived words reduced to their root forms using language-adapted stemming algorithms or expanded using linguistic ontologies for further classification purposes.
  • Fast profiling. Quick and easy overall notion of the text data content, providing basic information needed for more advanced text-mining tasks.
  • Named entity extraction. Extraction of real-life entities (people names, company names, countries, cities, phone numbers, etc.) mentioned in the text using either handcrafted rules or statistical methods.
  • Classification. Classification of data objects based on algorithms that take into account the classification probability based on the training data set and additional input variables, or based on clustering or specialized lookups.
  • Sentiment detection. Detection of the sentiment of input texts based on probabilistic methods enhanced with a subjectivity lexicon (set of common words that determine sentiment polarity and their weights) and rule-based optimizations.


Personalized marketing

Discover a particular client’s needs—finding the important keywords and phrases—and target the marketing campaigns accordingly.

Content categorization

Get a rough idea of what the given text is about, categorize it, and deal with it accordingly.

Churn prediction

Predict and prevent customer loss using customer’s mood detection, relationship analysis, or words indicating customers’ behavior.

Loyalty analysis

Find loyal customers and use their potential to increase the brand’s credit using sentiment analysis.

Customer feedback

Get customer feedback throughout multiple channels, including online discussions or social media threads.

Fraud detection

Identify financial fraud, predict and prevent loss caused by customer fraud employing keywords and language pattern spotting.

Risk analysis

Estimate customer’s credit and responsibility and acquire the most profitable clients while minimizing risk with significant expressions detection.

Competitive intelligence

Analyze external resources to understand what is happening outside the company’s business.