Skip to content

Solution

icon1

Gathering & Analytics

  • Analyze and Structure
  • Capture Data and Information
  • NLP stack

Capture from Multiple Sources

Internal and External Data

Define Relevant Data Capture Processes

icon2

Structuring and Prediction

  • Administration
  • Taxonomy Management
  • Prediction
  • NLP Layer
  • Build Knowledge Base

Processing / Analysis

NLP & Taxonomy Structuring & Predictive Reasoning

Measurable Values via Non-Linear Inference

Train System to Individual Topics

icon3

Interactive Presentation

  • Dashboard
  • Analysis & Drill Down

Present Results

User-friendly Dashboard

Searched Documents & Calculated Predictions

Implementation Timeline

01
Set up / Parametrize

1 - 2 days

02
First Operational Solution

3 - 4 weeks

03
Continuous Refinement of Solution

Ongoing

Infographic

The Data Loading and Data Preparation Layer handles all data loading and document pre-processing activities.

Data capture is supported from three different sources: internal data, social media screening and web-data crawling/RSS feeds.
During the data loading process, each loaded document is processed via a complete NLP (Natural Language Processing) stack consisting of:
  • Document transformation from any format to pure text format
  • Each document is split into individual pages for easier topic location within large text documents (for pdf documents only)
  • To improve the search capability a complete Indexation of each word per document is performed
For supporting document analytics the NLP stack further includes:
  • Sentiment analysis for document content
  • Tokenisation for word and sentence splitting
  • Lemmatisation (grouping together the inflected forms of a word so they can be analysed as a single item)
  • Stemming (process of reducing inflected (or sometimes derived) words to their word stem, base or root form)
  • Tagging and keyword identification
  • NER (Named Entity Recognition) for identifying names and companies
  • POS (Part of Speech) identification for semantic text analysis

The Learning and Knowledge Capture Layer consists of all services required to train the text analytics models (NLP) to individual knowledge domains, to build individual taxonomy and ontology structures for semantic content categorisation and to build predictive reasoning models using Dydon’s NeuroFuzzy concept.

The Presentation Layer has all generated results presented using parameterizable, easy to use and intuitive dashboards.

The Process

graphic2

Dydon Technology & Architecture

Java core base with open-source components

Database: MongoDB, NoSQL & REST APIs​

AI modules: Apache Lucene/Solr, OpenNLP, Mallet, Fuzzy4j​

Fully web-based, cloud and on-premise operation​

NLP Stack Components

Graphic

Get in Touch

  • Hechlenberg 17 CH-8704 Herrliberg
  • +41 79 457 87 67
  • info@dydon.net

or Schedule a Demo

We won't waste your time