|
PlainKnowledge For WinDream
Linguistic Enabling
Windream DMS
1. Sector / Application
PlainKnowledge for Windream expands Windream into a knowledge management
solution for
small, medium, and large enterprises. The system provides automated data
analysis and
classification to extract keywords from unstructured texts and sort
documents into pre-defined
categories.
2. Description of functions
The knowledge management software PlainKnowledge makes the information
streams within a
company manageable by automatically and systematically analyzing and
classifying knowledge
in text documents. Accessibility is increased significantly.
Knowledge is the basis of modern information driven enterprises. The ability
of a company to
optimally deploy its employees' knowledge becomes increasingly important and
difficult as
information density increases. Although knowledge is contained in documents,
texts, and
correspondence, it is frequently not used to its maximum effect because
employees aren’t
aware of critical knowledge at key decision points. Increasingly, a lot of
time is wasted
searching for information that can’t be located. Computer driven knowledge
management
through PlainKnowledge, developed by
AppTek, assists companies in organizing information
and putting it to better use:
 •
Analysis and categorization of
unstructured documents
• Key word extraction
• Categorization of unknown documents
in theme specific categories
• Sorting of texts into theme specific
folders (e.g. emails or ticker messages)
• Search existing data by analyzing document content and automatically
categorizing it, the processing and sorting of large amounts of text data
such
as emails or ticker messages is modularized and made more efficient.
The technological basis for
PlainKnowledge lies in statistical text analysis. This obviates the need for
developing complex rule-based systems that are difficult to adapt when the
environment changes. Statistical text-analysis is not only easy to use and
adapt to individual needs; it is also proven to produce better
classification results. The idea behind statistical text analysis is that a
machine can learn semantic connections from example texts. This simulates
the way humans understand text in context and is applied to automated
categorization.
The adaptability of PlainKnowledge for Windream is particularly true with
regard to choice of
language. Generally, the language of categorized documents is only dependent
on the language
of texts provided for the system to learn from. Classification can therefore
be applied to any
language. In order to improve quality, word lists are created that have
little or no meaning for
the specific language being used. PlainKnowledge is delivered with standard
wordlists for
German and English and further comes with an individually editable
stop-word-list from
Windream. Further languages are delivered upon request! The current version
(1.3) of AppTek’s
knowledge management packet contains the following modules:
• PlainClassify - Training
Company specific training of the categorization principle
• PLainClassify
Automatic semantic indexing of unstructured documents
• PlainCluster
Automatic data analysis and keyword extraction
• PlainRecherche
Category specific searching
• PlainIndex
Keywords as semantic indices
Classification functions can be used to automatically forward email, to
conduct efficient metaindex searches, or to maintain consistency in a
company filing system. Further, the company is currently working on
integrating associative searching (PlainRetrieve), which will complement
Windream’s full text searching function with a semantically based search
function to iteratively canvas a database.
3. Description of the Windream integration
The seamless integration of PlainKnowledge for Windream into the Windream
interface is not
only practical; it also creates economically valuable advantages:
• Minimal training time for employees/users
• Synergy effects (e.g., full-text extraction, version control of
categorized documents,
communal use of stop-word-list…) of knowledge management and document
management.
• Categorization of documents of all types from which Windream can extract
text, e.g.
DOC, HTML, PPT, etc.
• Concurrent manning of automatically and individually developed indications
• Use of current research in statistical classification for a low add-on
cost.
|
|
