important problems involve clustering large datasets. Although naive implementations of clustering are computa- tionally expensive, there are established ecient techniques for clustering when the dataset has either (1) a limited num- ber of clusters, (2) a low feature dimensionality, or (3) a small number of data points. However, there has been much less work on methods of eciently clustering datasets that are large in all three ways at once|for example, having millions of data points that exist in many thousands of di- mensions representing many thousands of clusters. We present a new technique for clustering these large, high- dimensional datasets. The key idea involves using a cheap, approximate distance measure to eciently divide the data into overlapping subsets we call canopies .T hen cluster- ing is performed by measuring exact distances only between points that occur in a common canopy. Using canopies, large clustering problems that were formerly impossible become practical. Under reasonable assumptions about the cheap distance metric, this reduction in computational cost comes without any loss in clustering accuracy. Canopies can be applied to many domains and used with a variety of cluster- ing approaches, including Greedy Agglomerative Clustering, K-means and Expectation-Maximization. We present ex- perimental results on grouping bibliographic citations from the reference sections of research papers. Here the canopy approach reduces computation time over a traditional clus- tering approach by more than an order of magnitude and decreases error in comparison to a previously used algorithm by 25%.

http://dc-pubs.dbs.uni-leipzig.de/files/McCallum2000Efficientclusteringofhighdimensionaldatasetswith.pdf

Efficient clustering of high-dimensional data sets with application to reference matching

A method and system for producing graphics. A hierarchical structure of a database is determined. A visual table, comprising a plurality of panes, is constructed by providing a specification that is in a language based on the hierarchical structure of the database. In some cases, this language can include fields that are in the database schema. The database is queried to retrieve a set of tuples in accordance with the specification. A subset of the set of tuples is associated with a pane in the plurality of panes.

Computer systems and methods for the query and visualization of multidimensional databases

A transformation description language (TDL) for specifying how data is to be manipulated in a data warehousing application. The TDL is comprised of a source for storing raw data, one or more transformation objects for processing the raw data according to predefined instructions, and a target for storing the processed data. A mapping is used for directing the data flow between the I/O ports corresponding to the source, the plurality of transformation objects, and the target. The mapping specifies the connectivity between the source, transformation, and target objects as well as the order of these connections. There are a number of different transformations which can be performed to manipulate the data. Some such transformations include: an aggregator transformation, an expression transformation, a filter transformation, a lookup transformation, a query transformation, a sequence transformation, a stored procedure transformation, and an update strategy transformation.

Apparatus and method for performing data transformations in data warehousing

A system and method for collecting, filtering, analyzing, distributing and effectively utilizing highly relevant events (such as key business events) in real time, from huge quantities of data. The present invention analyzes both historic and real-time data stemming from operational activity, by interfacing with internal data repositories (such as Enterprise Resource Planning (ERP) and Customer Relationship Management (CRM)), external data sources (such as suppliers and clients), and real time operational systems in order to create an Active Intelligence Platform. This Active Intelligence Platform is positioned as a layer between the organization's data sources and its applications, monitoring inputs and relaying only the important data items to the relevant individuals and/or systems. This allows individuals and systems to respond immediately and effectively to key events.

System and method for analyzing and utilizing data, by executing complex analytical models in real time

Generation of output or reports on information contained in a data source which may be any of two or more types of source data, in a standardized or uniform manner is provided. A plurality of drivers are provided specific to different types of source data which include programming for identifying structural or other characteristics of the various data sources, e.g. for use in defining a new database. Preferably the new database is configured to permit highly flexible and/or rapid output or reporting or is otherwise optimized for reporting purposes. In one embodiment, the present invention includes conversion of one or more data sources into one or more uniform databases, preferably generating one or more key categories for organizing the data, optionally generating category groupings or rollups and additional data or optional references. In one embodiment, one or more databases are created which have a degree of uniformity of structure, even though they may be based on two or more different data sources which may have very different structures. The different data sources are automatically analyzed and this analysis can be used to identify and/or create categories of data for use in organizing the database.

Data retrieval method and apparatus with multiple source capability

A computer software architecture to automatically optimize the throughput of the data extraction/transformation/loading (ETL) process in data warehousing applications. This architecture has a componentized aspect and a pipeline-based aspect. The componentized aspect refers to the fact that every transformation used in this architecture is built up with transformation components selected from an extensible set of transformation components. Besides simplifying source code maintenance and adjustment for the data warehouse users, these transformation components also provide these users the building blocks to effectively construct pertinent and functionally sophisticated transformations in a pipelined manner. Within a pipeline, each transformation component automatically stages or streams its data to optimize ETL throughput. Furthermore, each transformation either pushes data to another transformation component, pulls data from another transformation component, or performs a push/pull operation on the data. Thereby, the pipelining; staging/streaming; and pushing/pulling features of the transformation components effectively optimizes the throughput of the ETL process.

Method and architecture for automated optimization of etl throughput in data warehousing applications

A method of performing incremental aggregation of dynamically increasing database data sets. An embodiment of the present invention operates within a data mart or data warehouse to aggregate data stored within an operational database corresponding to newly received data to provide current information. Initially, a computer server creates an intermediate file which is initialized by the server with an aggregate data set. The aggregate data set consists of data values and count values that each correspond to specific group identifiers. The computer determines if any group identifiers within a new set of inputs data are identical to any group identifiers stored within the intermediate file. If an inputted group identifier matches a stored group identifier, the inputted data value is aggregated with the stored data value and the count value corresponding to the specific stored group identifier is incremented by one. If an inputted group identifier does not match any of the stored group identifiers, the inputted group identifier and corresponding data value are stored within the intermediate file and a count value of one is appended to that specific group identifier. Once all the group identifiers within the new set of input data have been determined, the computer stores all the changes that were made to the intermediate file into the aggregate data set.

A method for incremental aggregation of dynamically increasing database data sets

A method for updating a target table of a data mart in response to changes made by a transaction to data stored in a source table of an operational database. Data that was changed in the source table by the transaction is stored in a dynamic image table of a change capture database. Data that was not changed in the source table by the transaction, but which is nevertheless required to be mapped to the target table, is stored in a static image table of the change capture database. The change capture database also contains relevant information regarding the transaction. Once the dynamic and static image tables are properly staged, the changes are propagated from the change capture database to the target tables of the data marts. In other words, data is extracted from the change capture database and subsequently transformed and loaded, thereby minimizing the impact to the operational database. Thereupon, the tables of the change capture database are truncated to discard data which is now no longer needed.

Apparatus and method for capturing and propagating changes from an operational database to data marts

A method and system thereof for performing real time transformations of dynamically increasing databases is described. A session, identified as a real time session, is initialized. The real time session repeatedly executes a persistent (e.g., continually running) data transport pipeline of the analytic application. The data transport pipeline extracts data from a changing database, transforms the data, and writes the transformed data to storage (e.g., a data warehouse or data mart). The data transport pipeline is executed at the end of each time interval in a plurality of contiguous time intervals occurring during the real time session. The data transport pipeline remains running after it is executed, until the real time session is completed. Accordingly, new data are transformed in a timely manner, and processing resources are not consumed by having to repeatedly re-establish (re-initialize) the data transport pipeline.

Mohan Sankaran

Papers

Method and architecture for automated optimization of etl throughput in data warehousing applications

A method for incremental aggregation of dynamically increasing database data sets

Apparatus and method for capturing and propagating changes from an operational database to data marts

Real time sessions in an analytic application

Update propagation to a data warehouses with specified latency