|
|
|
 Following the success of the demonstration sessions in previous KDD conferences, the KDD-97 program will also include demonstrations of knowledge discovery products, knowledge discovery applications and research prototypes. Unlike previous demonstration sessions, we will clearly differentiate between commercial product demonstrations and research demonstrations. |
 Both commercial product exhibits and research prototype demonstrations will be held at the Newport Beach Marriott Hotel and Tennis Club, Newport Beach, California, Friday, August 15, 1997, from 12.30pm to 5.00pm. |
 We invite commercial vendors to exhibit at KDD-97. The exhibitor fee for KDD-97 will be a nominal $250.00. Exhibitors will be provided with a 6 foot table top. In this space vendors will be allowed to distribute product or company literature, show product demonstrations and set up signage. Vendors will need to bring all necessary hardware and software that they will require for their demonstrations. The exhibit area will be open August 15th from 12:30-5:00 pm. Total attendance at KDD-96 was 457. Of these 35 percent were affiliated with universities and 65 percent were affiliated with industry. If you would like to exhibit at KDD-97 please fill out the registration form and send it along with the name of your product(s) and/or service(s) and a 200 word (maximum) description of product(s)/service(s) to: AAAI KDD-97 Exhibit 445 Burgess Drive Menlo Park, CA 94025 USA Your description will be published in the conference program. The current list of exhibitors is shown below. AcknoSoft
The KATE suite includes KATE datamining to generate automatically decision trees from cases, KATE-CBR to retrieve cases that are similar to a query and reason by analogy, KATE-editor to build a database that is object-oriented or link to most existing market databases. As an option, users may add KATE for R/3 Service Management to link to SAP and KATE WebServer to access the decision support system over an intranet or the internet. Written in C, KATE is also available as dynamic link libraries in Windows or as shared libraries on Unix (Sun Solaris). The KATE suite is tailor made for analyzing complex data in technical domains. Applications of KATE include: maintenance of Boeing 737 engines (Cfm), of marine diesel engines (New Sulzer Diesel, Switzerland), of Naples' metro (Ansaldo in Italy), of the French high speed train TGV (GEC Alsthom), help desk for robots in the plastic industry (Sepro in France), telecom networks (Alcatel in France), CAD/CAM workstations (Mercedes, Germany), reliability analysis of offshore platforms (Nork hydro Norway), nuclear power plants (French electricity), industrial gas meters (French Gas), the Ariane space center (Matra), quality management in manufacturing (Schlumberger), evaluation of costs for manufacturing plastic parts (Legrand, France), sales support for electronic devices (Analog devices). IBM Corporation
Researchers at many IBM labs around the world are continuously developing powerful algorithms for analyzing large data sets stored in databases or flat files. These algorithms cover a full spectrum of data mining technologies and enable analyses ranging from classification and predictive modeling to association discovery and database segmentation. They can run on small workstations but are highly scaleable since they have parallel implementations optimized to handle large and parallel super computers and databases. Along with a comprehensive set of data analysts and application developers through IBM's flagship data mining technology product, the Intelligent Miner. Leveraging the Intelligent Miner technology, IBM has developed a collection of applications, Business Discovery Solutions (BDS), to make data mining more accessible to Business Users. Through an easy to use Java GUI, BDS addresses business problems such as customer retention, risk analysis, store layout optimization, and cross selling. IBM's top-caliber data mining analysts have extensive industry expertise and have helped more than 60 companies exploit the new developments in data mining. IBM, on whose machines 70% of the world's data reside, supports all the components required to guarantee a customer's success in data mining. Information Discovery, Inc.
The Data Mining SuiteTM is a comprehensive and integrated set of data mining products that provide complete solutions for knowledge discovery, predictive modeling and internet/intranet-based applications within a unified frame-work. Each product is novel and useful in its own right, but the joint application of the techniques used within the suite delivers unprecedented benefits to corporate users. The Data Mining SuiteTM is highly scaleable and accesses large SQL databases directly without sampling or extracts. Kluwer Academic Publishers
Kluwer Publishers will have the journal Data Mining and Knowledge Discovery (Editors-in-chief: Usama Fayyad, Heikki Mannila, and Gregory Shapiro-Piatetsky) on display! Many other fine journals are also available for review, as well as over 25 new books, discounted 20% for KDD '97 attendees! MathSoft, Inc.
MathSoft, Inc., is a leading provider of knowledge discovery software, with more than a million users worldwide. Mathsoft has a flexible family of products, offering solutions for data analysis, statistical data mining and decision support. S-Plus is the premier solution for powerful data analysis, visualization and statistical data mining. S-Plus offers the richest data analysis environment available, with over 2,000 built-in functions including both classical and robust techniques, all within a customizable, intuitive user interface. In addition, exclusive TRELLIS graphics allow you to reveal hidden meanings in complex, multidimensional data. MathSoft Stat Server is a powerful new approach that delivers a competitive edge in decision support. The first enterprise-wide solution for distributing sophisticated analyses and graphics, StatServer leverages your company's existing client/server and Internet/intranet technology to put information in the hands of decision makers. Stop by to see a demonstration of the power of S-Plus and StatServer. Morgan Kaufmann Publishers
Since 1984, Morgan Kaufmann has published the finest technical information resources for computer and engineering professionals. Our audience includes the research and development communities, information technology (IS/IT) managers, and students in professional degree programs. We publish in book and digital form in such areas as databases, computer networking, computer systems, human computer interaction, computer graphics, multimedia information and systems, artificial intelligence, and software engineering. Many of our books are considered to be the definitive works in their fields. Please stop by our display and receive a 15% Knowledge Discovery and Data Mining 1997 conference discount. PC AI Magazine
PC AI Magazine provides the information necessary to help managers, programmers, executives, and other professionals understand the quickly unfolding realm of artificial intelligence (AI) and intelligent applications (IA). PC AI addresses the entire range of personal computers including the Mac, IBM PC, neXT, Apollo, and more. PC AI features developments in expert systems, neural networks, object oriented development, and all other areas of artificial intelligence. Feature articles, product reviews, real-world application stories, and a Buyer's Guide present a wide range of topics in each issue. Salford Systems
Salford Systems is exhibiting CART 3.0 (Classification and Regression Trees), a tree-structured data-mining tool, co-developed with the original authors of CART at UC Berkeley and Stanford (Breiman, Freidman, Olshen, and Stone). The premier decision tree tool-complete with built-in n-fold cross-validation, user-definable variable misclassification costs, linear combination splits, efficient handling of high-dimensional categorical predictors, and the new feature of combining multiple trees-is now available in one affordable package. CART 3.0 has a completely revised Windows graphical user interface and the ability to interact with the tree after database analysis. Useful diagnostics include dynamic pruning, gains charts for all sub-trees, and simultaneous viewing of training and test data accuracy scores. Experienced users can control CART 3.0 via command scripts, while newcomers can use point-and-click menu selections. Throughout an interactive session using the menu, CART 3.0 records command equivalents, providing an audit trail for the session. An in-depth, comprehensive manual explains every feature and nuance of CART within the context of over 30 examples. Salford Systems has been developing advanced tools for data analysis for PC and Unix platforms since 1983, and also provides consulting services to the telecommunications, financial services, health care and direct mail industries. Silicon Graphics
MineSet (TM) version 2.0 is the fourth release of SGI's product for exploratory data analysis. Combining powerful integrated, interactive tools for data access and transformation, data mining, and visual data mining, MineSet provides you with a revolutionary paradigm for getting maximum value from your vast data resources. MineSet enables you to gain a deeper, intuitive understanding of your data, by helping you to discover hidden patterns, important trends and new knowledge. It is this deep understanding which can be used for developing powerful business strategies leading to greater competitive advantage. SRA International
SRA International has been creating innovative solutions to practical problems faced by businesses and government agencies for over eighteen years. We specialize in the fields of intelligent information retrieval; machine learning; knowledge-based systems; database engineering; and natural language processing. SRA empowers organizations with the ability to discover and detect patterns critical to their success through the use of a complete line of scaleable data mining tools and professional services. SRA's KDD Toolset includes multi-strategy algorithms for discovering Associations, Classifications, Sequences, and Clusters, as well as high-speed rule and sequence-based pattern matching algorithms. These algorithms employ direct database access for mining data. Additionally, they are parallelized to take advantage of multiprocessor platforms for rapid analysis of extremely large data sets. Finally, we employ a comprehensive set of JDBC-compliant Java-based user interfaces for configuration and execution of algorithms as well as visualization of results for analysis and interpretation. SRA's knowledge discovery specialists understand how best to apply these advanced capabilities to enable you to utilize your most strategic asset: electronic information. Together, SRA's KDD Toolset and professional services provide solutions that are ideal for tackling such large-scale problems as fraud detection and prevention, competitive intelligence, and market behavior. Torrent Systems Inc.
Torrent Systems, Inc. develops and markets tools, component software and applications to support system integrators and applications developers in building advanced data warehousing and data mining applications that run on parallel processing systems. ORCHESTRATE, Torrentis parallel development environment, hides the complexity of parallel programming and facilitates the creation of fully parallel, high-performance data-processing solutions for your MPP or SMP systems. ORCHESTRATE is fully compatible with all major scalable servers including the IBM RS/6000, SP server using IBM DB2, Parallel Sysplex, Sun, HP, NCR, Digital and Intel and supports Oracle Parallel Server, Informix XPS, and IBM DB2/PE. Torrent Systems is headquartered in Cambridge, MA. Toshiba Corporation
NNE is a neural network based data mining tool. NEX (Neural network EXplainer) is the explanation module of NNE. NEX provides the explanation for trained neural networks by extracting rules from the networks. Since trained neural networks are black boxes and are difficult for humans to understand, the neural network is incomplete as a technique of data mining, which aims to discover understandable knowledge from databases. So NEX makes the neural network a complete data mining technique. NEX has the following features. (1) NEX can be applied to any neural network including recurrent neural networks. (2) NEX can be applied to any training method. (3) NEX can be applied not only to discrete values but also to continuous values. (4) NEX extracts accurate and simple rules in a short time. Especially, when classes are continuous, there is no other systematic method which can discover understandable knowledge. NEX discovers understandable knowledge, that is, rules, from trained neural networks. The accuracies of the rules are based on the trained networks. NEX can be connected to any neural network. So any neural network user can obtain an explanation for trained neural networks just by connecting NEX to their neural networks. WizSoft Ltd.
Wiz Why for windows 95/windows NT is a data mining application for issuing prediction and classification. WizWhy analyzes the data, reveals all the if-then rules and mathematical formula rules, and calculates the significance level of each rule. WizWhy then predicts future cases based on the discovered rules. In empirical tests WizWhy was found to be faster and more accurate than neural networks, decision trees and genetic algorithms. WizRule for Windows 95/windows NT is a data cleansing and auditing tool. WizRule reveals all if-then rules and formula rules in the database. It then points at the deviations from the set of all the discovered rules as suspected errors, and calculates the level of unlikelihood of each deviation. WizRule avoids false alarms; almost every deviation with a high level of unlikelihood is indeed an error. |
 We are also soliciting demonstrations of research prototypes at KDD-97. This demonstration session will be held on August 15 from 12:30 to 5:00 pm. We have a limited budget for providing hardware for research demonstrations. This year we will give priority to demonstrations that are in conjunction with accepted papers at KDD-97. Within budget and space constraints we will make every effort to accommodate as many demonstrations as possible. If you would like your demonstration to be considered for KDD-97 please provide the following information to Tej Anand (tej.anand@atlantaga.ncr.com) by June 1, 1997:
The current list of demonstrations is shown below. An Interactive Visualization Environment for Data ExplorationPaper Title:An Interactive Visualization Environment for Data Exploration
We will demonstrate an information-centric interface architecture for unifying the subtasks of knowledge discovery, allowing the analyst to focus on the process rather than the tools. These subtasks include datacleaning, creating a dataset, data reduction and projection, and exploratory visualization. Architectural integration is achieved with a shared object-oriented database, which is accessible to the user via a visual query language. Tight integration between queries and visualizations make exploration more interactive and less of a feed forward process than in previous systems. What Is Unique about the System? Document ExplorerPaper Titles:(1) Visualization Techniques to Explore Data
Mining Results for Document Collections, and (2) Maximal Association Rules: A
New Tool for Mining for Keyword Co-Occurrences in Document Collections
Document Explorer is a data mining system for document collections. Such a collection represents an application domain, and the primary goal of the system is to derive patterns that provide knowledge about this domain. Additionally, the derived patterns can be used to browse the collection. Document Explorer searches for patterns that capture relations between concepts of the domain. The patterns which have been verified as interesting are structured and presented in a visual user interface allowing the user to operate on the results to refine and redirect mining queries or to access the associated documents. The system offers preprocessing tools to construct or refine a knowledge base of domain concepts and to create an intermediate representation of the document collection that will be used by all subsequent data mining operations. The main pattern types the system can search for are frequent sets, associations, concept distributions, and keyword graphs. To enable to provide some explicit bias, the system provides a dedicated query language for searching the vast implicit spaces of pattern instances that exist in the collection. This query language offers syntactical, background, quality and redundancy constraints. The query language is embedded in a GUI which makes it easy even for novice users to explore the document collections. What Is Unique about the System? GeoMiner: A Geo-Spatial Data Mining EnginePaper Title:Described in Proceedings of SIGMOD'97
Spatial data mining is to mine high-level spatial information and knowledge from large spatial databases. A spatial data mining system prototype, GeoMiner, has been designed and developed based on our years of experience in the research and development of relational data mining system, DBMiner, and our research into spatial data mining. The data mining power of GeoMiner includes five spatial data mining modules: characterization, comparison, association, clustering, and classification. The SAND (Spatial And Nonspatial Data) architecture is applied in the modeling of spatial databases, whereas GeoMiner includes the spatial data cube construction module, spatial on-line analytical processing (OLAP) module, and spatial data mining modules. A spatial data mining language, GMQL (Geo-Mining Query Language), is designed and implemented as an extension to Spatial SQL, for spatial data mining. Moreover, an interactive, user-friendly data mining interface is constructed and tools are implemented for visualization of discovered spatial knowledge. What Is Unique about the System? Id-VisPaper Title:A Visual Interactive Framework for Attribute
Discretization
We will demonstrate the Discretizer module of Id-Vis, our interactive platform for visual data mining on a client-server architecture. This module features multiple available algorithms, a drag-and-drop cut-point editor, multiple levels of data visualization with drill-down capability etc. A number of the variables are exposed to user experimentation. The system provides visual cues to the "optimal" number and locations of the cut-points. It also provides feedback to the user about the extra "badness," over the system-derived optima, incurred during the experimentation. The central philosophy is that the system should place the user within the appropriate context of system-derived values, and provide the user with the opportunity to intelligently modify (e.g. display the impact on accuracy of these modifications) these optima and propagate them downstream. The server originally ships a compacted version of the raw data, or a equi-probabilistically distilled density function to the client. If, during the drill-down process, the user's information request and the associated accuracy constraints cannot be satisfied by the locally available data, the client obtains the appropriate data from the server and displays it. What Is Unique about the System? Interactive Knowledge Exploration Using DBMinerPaper Title:Metarule-Guided Mining of Multi-Dimensional
Association Rules Using Data Cubes
With years of research and development efforts, the DBminer system developed in CS/SFU, Canada has incorporated many advanced research results into our system. This includes multiple-level knowledge in large relational databases and data warehouses, a wide spectrum of data mining functions, including characterization, comparison, association, classification, prediction, and clustering, and a data visualization package. The major technologies adopted are integration with data warehouse and OLAP technology, attribute-oriented induction, statistical analysis, progressive deepening for mining multiple- level knowledge, and meta-rule guided mining. The system provides a user- friendly, interactive data mining environment with good performance. What Is Unique about the System? Kensington - High-Performance Distributed and Parallel Data MiningPaper Title:Large Scale Data Mining: Challenges and Responses
Kensington is a prototype of an open, distributed, web-based, high-performance data mining system for use on parallel servers. A web-based client-tool, written in Java, gives access to a distributed collection of data bases and data mining modules which are executed on a high-performance parallel machine. The application consists of components which are integrated using Java/Corba middleware. The main components are: a database server, a data mining server, a visualisation and Web-server, and the client control tool. The data mining servers are embedded in Corba objects and distributed across a LAN or WAN. The database server is accessed via JDBC. The client-tool can access and control the data mining actions from anywhere. The high-performance mining modules are portable parallel implementations in C and MPI, and cover commonly needed functions such as classification (C4.5), prediction (neural networks), association discovery and self-organizing maps. The light-weight client tool is enriched by visualization applets for data mining results, whenever they become available. The overall goal is to (a) integrate distributed servers, e.g. data and computation servers, and (b) to make the integrated system universally accessible over the Web. What Is Unique about the System? Mining For Many Kinds of KnowledgePaper Title:Knowledge = Concepts: A Harmful Equation
We will demonstrate Forty-Niner (49er), an automated discovery system that discovers knowledge in databases. We will apply the system to several databases to demonstrate how different forms of knowledge can be automatically discovered. 49er searches for many types of knowledge. It starts from contingency tables (CTs) and then recognizes special types of CTs which lead to other, more specialized forms of knowledge, such as equations, equivalence relations, or subset relations. When many relations of the same type have been discovered, 49er combines them into forms such as taxonomies and subset graphs. We will contrast 49er with specialized systems that are focused on a single form of knowledge, even if other forms of knowledge are much more appropriated for a given dataset. Further, we will contrast 49er's focus on knowledge that contains as much of empirical contents as possible with the focus on concept definitions typical to machine learning. We will show how CTs can be used to generate decision trees and how 49er's search for additional "redundant" knowledge makes decision trees more flexible and statistically significant. We will also contrast 49er's approach to taxonomy formation (combine many approximate equivalence relations) with statistical and conceptual clustering. What Is Unique about the System? SONAR (System for Optimized Numeric Association Rules)Paper Title:Computing Optimized Rectilinear Regions for Association Rules
Recent progress in technologies for data input have made it easier for finance and retail organizations to collect massive amounts of data and to store them on disk at a low cost. Such organizations are interested in extracting from these huge databases previously unnoticed information that inspires new marketing strategies. In this demonstration, we introduce a system for mining optimized association rules and for generating decision/regression trees from databases with numeric data as well as categorical data. What Is Unique about the System? S-PLUS DataBlade for Informix Universal ServerPaper Title:Data Mining with
Trellis Graphics
Informix Universal Server (IUS) is an object-relational data base. S-PLUS is an object-oriented language and system for data analysis, statistical modeling, visualization and programming with data. An IUS DataBlade is a collection of types (classes) of objects and access methods, that closely integrates applications software with the IUS database, typically on the server. The S-PLUS datablade for IUS provides new data types in IUS corresponding to intrinsic S-PLUS datatypes and functions to apply any S-PLUS expression on these datatypes. It also provides functions to convert IUS native data to these S-PLUS datatypes and vice-versa. The demonstration will include several data mining applications examples, including: integrated query and statistical data mining, visualizing and modeling the relationship between equity returns, firm size and book-to-market; robust beta mining (finding firms listed on the AMEX, NASDAQ and NYSE or which the beta calculation is influenced by outliers, and visualizing the data for such firms); hexagonal binding visualization of scatterplots for largish data sets; application of trellis graphics to a Lucent customer value analysis (CVA) study. What Is Unique about the System? |
home | top |