i2k AI Process and Portal

i2k maximizes AI technology and natural language processing to drill into a company's rich sources of knowledge.

From the beginning of implementation, the i2k AI Platform delivers economic value by reframing pieces of disparate data into intelligent, usable content.

Give i2k Connect a pile of documents and the i2k AI Platform immediately goes to work applying built-in knowledge related to your working domain. As that happens, the i2k Connect team guides your team through the transformation from manual searches for information to reaching personalized search results by reframing pieces of disparate data into intelligent, recognizable content sets.

DeliveryAI PipelineDiscoveryi2k AI Platform(click the icons below to view more information)

i2k AI portal

Once launched, the i2k AI portal includes an optional four-panel dashboard that provides overview data for the collection of records returned by a user’s search:

  • a timeline of documents and their chronological frequency
  • a concept tag cloud based on important words or phrases
  • an interactive map of geopolitical locations identified in the content
  • a social network graph of collaboration among authors and institutions.
File overview

The overview for each file shows

  • a thumbnail image
  • basic metadata such as source and date
  • additional automatically determined descriptive tags such as title, summary, concept tags, and classifications into taxonomic classes of interest.

All tags make user review quicker and easier, saving significant time in research, as much as 90% less time spent finding and analyzing data.

Classifications

In the out-of-the-box software, you can select from a hierarchical list of about 6,000 classes arranged in 17 taxonomies. Each of the taxonomies indicates a broad category, perspective, or “view” that is available to refine a search. The i2k Platform can also classify according to customer-supplied taxonomies.

Concept tags

Documents are automatically tagged with unique, accurate, and consistent metadata using AI technology, informed by industry knowledge.

Structured

Information that is organized in a standard format and easy to define and search. It can be arranged by columns and rows, in a database. About 20% of an organization’s data is usually structured.

Unstructured

Any type of file that does not have a predefined structure or format and can be part of a data lake. It cannot be represented in columns, rows, or databases. Examples include PDF, PowerPoint, Word, images, and social media data. About 80% of an organization’s data is unstructured.

Graphical

Can include images or video data.

Text Miner

Extracts the text and text structure from a file.

Headliner

Identifies the title of a file, distinct from its name.

Summarizer

Delivers a summary of the content of a file.

Classifier

Categorizes a file into relevant taxonomies.

Identifier

Tags concepts, places, companies, and other entities mentioned in files.

Table Miner

Detects and interprets tables in PDF files.

SQL Miner

Extracts and summarizes data points from SQL databases.

Data Miner

Extracts and summarizes data points from unstructured files.

Analyzer

Summarizes metadata distributions and trends for sets of files, including near duplicates.

Portal

Offers integrated search, browsing, analysis, and visualization.

Newsflash

Delivers alerts when updates essential to your work are published.

Query

Conveys intelligent answers to user questions.

Integrator

Delivers enriched metadata to your existing information management systems.

Web API

An application programming interface for the Web