March 9-10, 1999
McLean Hilton
7920 Jones Branch Drive
McLean, Virginia
Megaputer will be attending the 1999 Federal Data Mining Symposium & Exposition, presented by AFCEA International and participating Federal Agencies. The major goals of Federal Data Mining ‘99 include:
- Identify and discuss new data mining techniques
- Exchange information and ideas on the role of data mining
- Present requirements and proposed solutions
- Provide policy guidance applicable to DoD and civil agency information resource managers
- Identify and encourage service-unique and government-wide knowledge
AFCEA joins the Data Mining Committee in its determination that this symposium will provide a singular opportunity for its participants to share in the achievement of these goals.
Tools that Megaputer will be exhibiting:
- PolyAnalyst – multi-strategy data mining suite
- TextAnalyst – semantic text analysis and information management
- PolyAnalyst COM – simple development kit for DSS applications
- CrossReader – internet surveying on demand
PolyAnalyst
Multi-strategy data mining solution utilizing the latest achievements in knowledge discovery in databases. Works with numeric, categorical, and yes/no variables. Exploration engines predict values of continuous variables, solve classification and clustering tasks, determine the most influential independent variables, and explicitly model complex phenomena. Statistical significance of models is rigorously checked through a novel randomized testing technique.
PolyAnalyst furnishes numerous machine learning algorithms:
- Find Laws (Symbolic Knowledge Acquisition Technology – SKAT)
- PolyNet Predictor (neural net and GMDH hybrid)
- Cluster (Localization of Anomalies)
- Find Dependencies (Multidimensional distribution analysis)
- Classify (Fuzzy logic classification)
- Discriminate
- Stepwise Linear Regression
Provides direct access to data held in ODBC-compliant databases. Data and exploration results can be exchanged with MS Excel, and visualized in numerous formats: histograms, line and point plots with zoom and drill-through capabilities, colored charts for three dimensions, interactive Rule-Graphs with sliders for effective presentation of multidimensional relations, Lift and Gain charts for marketing applications, and a special Frequencies graph for a thorough visualization of the distribution of categorical, integer, and yes/no variables.
Point-and-click GUI and versatile data manipulation, visualization, and reporting capabilities make PolyAnalyst a very easy-to-use system.
TextAnalyst
Unique intelligent tool for automated semantic analysis, summarization, and navigation of unstructured natural language texts. Can work with a single text or huge textbases. Performs a broad range of text mining tasks:
- Semantic network development for a corpus of analyzed texts
- Text abstracting
- Knowledge base creation
- Semantic search for information (with the creation of a sub-tree of concepts related to the concept mentioned in the query)
- Hierarchical topic structuring
- Single-click text navigation
- Concept tagging
TextAnalyst utilizes an innovative neural network technology for a homogeneous processing of texts from arbitrary fields. In combination with other implemented advanced text processing techniques this approach provides for the most efficient development of the semantic network of a text. The developed accurate semantic network is used by TextAnalyst to perform further multi-faceted analysis of texts.
Existing users of TextAnalyst include government offices, consulting and law firms, medical centers, scientific organizations, electronic book publishers, customer support centers, political institutions, and even college students.
PolyAnalyst COM
A toolkit of COM objects for incorporating world’s most powerful data mining algorithms in arbitrary vertical applications. Allows creation of advanced decision support systems within a few days utilizing simple development tools like Visual Basic. Includes well documented description of all interfaces.
PolyAnalyst COM is beneficial to users who want to:
- pay only for those components which are needed
- keep using his/her favorite application interfaces, but access new functionality
- readily exchange information between different applications
- run applications on a remote computer
- use distributed data analysis applications
COM modules are readily incorporated in other applications, like Matlab or Microsoft Office products, with absolutely no programming involved. Powerful modeling capabilities of PolyAnalyst can be used, while working with a familiar interface of a calling application. In addition to enhanced data manipulation, visualization and report generating capabilities, the following machine learning algorithms are implemented in PolyAnalyst COM:
- Memory Based Reasoning
- PolyNet Predictor
- Cluster
- Find Laws
- Find Dependencies
- Classify
- Discriminate
- Stepwise Linear Regression
CrossReader
Intelligent search, text mining, and knowledge management solution. Creates on-demand surveys of analyzed texts. Represents a unique value to common internet users as well as to professional analysts engaged in processing large volumes of unstructured textual information. Analyzes information either stored in a local textbase or found in documents at specified remote web sites. Can perform an open-ended surveying powered by Internet search engines.
CrossReader main features include:
- searching for text excerpts containing the query words or phrases, or encountered in a certain context
- suggesting additional keywords relevant to the subject of interest, based on semantic analysis of initially found excerpts, and performing a new search with a shifted focus
- assembling excerpts in a text-like mode, with neighboring text fragments sharing common semantic elements – an on-demand survey
The user can visually adjust the search mode settings determining the amount of relevant information retrieved: from a digest compiling a few most important fragments, to a survey including a wider range of excerpts more subtly related to the initial query.