ACUMI DataFind 2018-02-01T01:33:49+00:00

ACUMI DataFind

Text analytics

ACUMI DataFind uses predictive interpretations and decisions derived from unstructured data found in social media, PDF, voice-to-text, handwritten notes and other sources. It then combines unstructured data with structured data to produce true and complete insights.

Clients often begin with this text analytics solution to accurately interpret the unstructured data flowing
through their organizations.

Understanding unstructured data can have a dramatic impact. You can see results within a few weeks of engaging our data scientists to prove the ACUMI concept. A working prototype system, using your data, is quickly constructed with ACUMI DataFind, followed by a brief
pilot program. Production is deployed shortly thereafter. Our solutions are built for speed and high-quality, innovative results. We’ll prove it.

Since data is the foundation for repeatable, successful decisions, you can’t take rational actions when missing key information. ACUMI DataFind enables you to easily discover and add new data sources. This intelligent solution moves with your business while providing predictive insights that you can’t otherwise find in your data. ACUMI DataFind extracts the true value of your data, both structured and unstructured.

Key programs and functionality of ACUMI DataFind include:

Ontology Generator

This module automatically discovers relevant terms and expressions from written communications, such as chats, email and voice-to-text. It’s often difficult to capture customer-specific ontologies. Some are non-intuitive and even your best expert may miss them. Ontology Generator enables an analytical approach that doesn’t rely entirely on domain experts.

The Ontology Generator results are used to set up and train cognitive solutions. The module provides higher cognitive accuracy in a shorter timeframe, as it uses best-of-class algorithms to ensure relevant results.

Natural Language Processor (NLP)

This module pre-processes text to remove punctuation and mark-up “noise” for subsequent analysis, providing a cleaner view of the text for analysis and human annotation. The NLP reduces analyst effort, as it’s configurable for specific applications and data sources.

Concept Extractor

This module identifies and extracts the most critical concepts to your organization—with high precision. It’s critical to discover changing or emerging concepts. Concept Extractor automates the discovery process rather than looking for it after the fact. It captures your knowledge workers’ expertise and applies it to the data, resulting in the remarkable ability to scale top staff member expertise through automation.

Attaining these concepts from your knowledge workers is the game-changing value of automating through AI. Across industries, Concept Extractor captures analytical best practices, including clinical, legal, regulatory, customer service and administrative. The information supports annotations in cognitive solutions. Benefits include scaling human knowledge and expertise, increasing productivity, reducing errors, improving risk management and lowering liabilities.

Scaling best practices has become increasingly impractical for staff, especially with the proliferation and production of structured and unstructured data. Concept Extractor covers financial issues, customer identification, clinical and common services. It’s highly configurable to process even brand-dependent concepts, organizational metrics, regulatory standards, services treatments, investigations, offerings and promotions. The module then applies these special skills and concepts to text analytics. This increases
accuracy of analysis, which translates into more complete process automation.


This module represents specialized classification algorithms curated and tuned to provide the best performance on customer conversations. The resulting models provide real-time processing by a cognitive solution or for analysis. The Classifier asset surpasses generic classification equivalents by providing robust models for skewed data sets common to customer service communications and investigative research. Classification enables you to learn from your experiences.