Codify • Summarize • Index
Abstract and codify concepts in text for knowledge
mining, document indexing, retrieval
and summary reporting
AutoCode is software that lets you automatically index, abstract and categorize information contained in unstructured text.
Digital documents contain valuable information, but finding what you need can be a challenge because it is
often embedded in the text. Consequently, if you want to find documents containing specific information you have to rely on an
index, or an abstract, to help with your search. However, if the index doesn't cover your area of interest, you won't find the
documents you need.
AutoCode lets you automatically index and codify information in text. AutoCode consists of a natural language
parser, an inference engine, and lexicons. AutoCode, with a specific lexicon, will search text for the concepts
contained in that lexicon.
Automated coding of cancer pathology reports is one application of AutoCode — other applications are also
possible, such as:
- Identification and coding of diseases and medical conditions
- Automated document indexing for future retrieval
- Abstracting information from medical records
- Mining text for information not previously indexed
Context sensitive parsing
What distinguishes AutoCode from other automated coding systems is its context sensitive parsing capability.
This allows AutoCode to not only identify concepts in text, but also the negation of concepts, something that simple keyword searches
cannot do. What's more, AutoCode handles variations in wording and ambiguous terminology. These features enable AutoCode to achieve very high
sensitivity and specificity scores when used as a document filter.
Lexicons
To automatically code text to a particular set of concepts, you need a lexicon. A lexicon encapsulates the concepts,
words and style of discourse within a particular subject area.
The Lexicon Manager desktop application is used to create and edit lexicons. Lexicons are built by entering all of the
concepts and terms for the subject area of interest and specifying any special language parsing rules that may apply. The Quick Test utility
is used to assess and tune the coding performance of the system.
Custom Solutions
AutoCode is a powerful tool for extracting meaningful information from text with many possible applications in
the areas of research, information retrieval, and document management. With its rule-based inference engine, AutoCode can also be used
to implement artificially intelligent information indexing and distribution systems.
If you have a particular need to index, categorize or search information contained in text, AIM's knowledge engineers
and developers can implement a custom solution for you. We can help you create the required lexicons, interface AutoCode with your databases
or file systems, and even embed AutoCode into your existing information systems.
Summary of AutoCode Features
- Natural language processing with negation detection
- Lexicon driven concept identification
- Handles ambiguous terminology, acronyms and abbreviations
- Fully integrated rule-based inference engine
- Lexicon Manager desktop application for Windows
- Quick Test desktop application for Windows
- HL7 and XML data import/export capability
- Integrates with AIM's TransMed Application Integration Engine
- AutoCode DLL API for embedded applications
AutoCode is a key element of AIM's E-path Technology.
|
