Implementation

Deployment and Installation

80% of Behavox customers chose to deploy in the cloud saving 75% on total cost of ownership.

Alex Wood Chief Commercial Officer, Behavox

The Behavox Platform can be installed on-premise or on the Virtual Private Cloud (VPC) hosted either by the customer, within the customer’s infrastructure, or by Behavox.

Virtual Private Cloud (VPC)

  • No capital infrastructure investment required.
  • VPC can be quickly scaled up or down without disruption to the service.
  • Behavox can maintain the hosted environment providing further cost savings.
  • Data is securely stored on VPC in compliance with data protection and retention regulations.

On-premise

  • All customer data is stored on-premise.
  • No dependency on external vendors for infrastructure and security.
  • Behavox can provide professional services support for integration and maintenance.
80% of Behavox customers chose to deploy in the cloud saving 75% on total cost of ownership.
Request architecture diagrams

Integration

Behavox delivers real-time integration of complex and diverse data sources.
Where customers have a proprietary data types, Behavox builds data connectors and assurance of data completeness at speed and affordable cost.
 
Critically, the Platform processes and analyzes structured and unstructured data.
datatypes.png
Ingesting, processing, cataloging and sorting of data is an automatic process with regular monitoring and reporting provided to the customer.
Request integration blueprints

Performance & Scale

Behavox Data Lake

Over 70% of Behavox customers chose The Behavox Platform as their primary Enterprise Data Lake. Behavox Platform is also compatible with customer's existing data lakes.
 
Aggregation of all Enterprise data on one storage and processing platform is often referred to in the industry as a ‘Data Lake’. Using the Behavox Platform as a Data Lake offers substantials benefits to the enterprise.
 

High performance

The Behavox Platform is designed to handle large and varied datasets, while deployed on a cluster of commodity hardware computers. This allows the Platform to achieve the same processing capacity as high-end supercomputers at a significantly lower cost.

Scalability

Highly distributed nature of the Behavox proprietary processing framework ensures that processing tasks get equally distributed between the nodes in the cluster in order to guarantee optimal load on each node. By adding extra nodes to the cluster and integrating them with the Hadoop Distributed File Storage system (HDFS), it is possible to linearly scale the available storage with no impact to its performance characteristics.

Compliance

All data is stored in accordance with data protection and retention regulations.
  • The Behavox Platform can be deployed in a specific region to comply with regional regulatory requirements for data storage and processing.
  • The Platform supports the setting of granular data retention policies based on different country, job function and data types. Retention periods can be set in days or years.
  • The Behavox Platform fully supports WORM storage with legal hold functionality to comply with SEC 17a-4 requirement.
  • Behavox ensures the ongoing confidentiality, integrity, availability and resilience of systems and services processing personal data.
  • Behavox uses machine learning to effectively mask or anonymize personal and sensitive data including social security and national insurance numbers.
Over 70% of Behavox customers chose The Behavox Platform as their primary Enterprise Data Lake.

Alex Glasman Chief Data Scientist, Behavox

zelov.jpg
Request architecture diagrams
Regulatory archiving case study

Machine Learning

The Behavox Platform reveals previously unidentifiable insights using a large collection of machine learning algorithms that are applied to both structured and unstructured data.
 
The Platform applies both supervised and unsupervised learning techniques to understand the content and context of people data and provide organisations with empirically tested inferences.
For instance, the Platform understands the meaning of text data by running part-of-speech analysis, entity recognition and sentiment analysis.
 
The Natural Language Processing (NLP) engine has been trained by the Behavox Data Science team to work out-of-the-box. The benefit of such an approach is immediate benefit (Return on Insight) from detection of the following entities:
  • Financial Instruments (tickers)
  • Bid/Ask spreads
  • Maturity
  • Call to action
  • Organisations and persons
  • Phone numbers and titles of contacts in signatures
  • Locations
  • Date and time
  • Money
  • Disclaimers
  • Research reports
To maintain the Platform security, Behavox machine learning classifiers can be retrained on customer environments without having to share or reveal underlying data.
 
Behavox data scientists are at the forefront of research on machine learning and often participate in academic conferences and publish papers.
Request NLP demo
Download Research Papers

Voice Recognition

Behavox users can search for keywords both in text and voice within the same interface.

Kiryl Trembovolski COO Behavox

Over the past five years, Behavox developed proprietary models that would give the Behavox Platform native capability to analyze voice files.
 
Today, the Behavox Platform is the only enterprise processing system that integrates voice, text and other data analytics to provide a seamless user experience.
 
Keyword Search (KWS) service allows to search a large predefined and configurable vocabulary of words and their location in audio recording.
 
The service is based on flexible architecture that allows modification of various components of the system. The core of the service is automatic speech recognition (ASR) system based on state-of-the-art Kaldi toolkit with Behavox proprietary extensions.
 
Instead of extracting a single most likely phrase, the Platform builds a connected graph (consensus network - CN) of all possible transcriptions with probability of each path. This allows the Behavox Platform to deliver more accurate results versus conventional speech-to-text approaches.
Request voice recognition demo
Download research paper