Unisys Corporation

Other

Jane Campbell Mazzagatti

Jane Van Keuren Claar

Steven L Rajcan

Tony Phan

Robert Buckwalter

Timothy William Collins

Charles Austin Parker

Duane J McCrory

Frederick W Reed

John C Waller

Dana M Zabilansky

In a KStore having K nodes with respective K node counts a method for updating the K node counts includes processing a K node in accordance with a received particle to provide a processed K node and updating a node count in response to the processing. The processing of the processed K node includes building the processed K node in response to the received particle. A node count of a beginning of sequence node is updated. A beginning of sequence node includes a beginning of thought node. The received particle can be a particle of sensor data having a corresponding sensor node to provide a sensor node determination. A node count of the corresponding sensor node is updated in response to the sensor node determination. A subcomponent node is determined in response to the sensor determination and a K node count of the subcomponent node is updated.

Method for processing K node count fields using an intensity variable

A data simulator receives a set of directives specified in a file and creates one or more datastreams from which a data structure may be built as specified by the directives. The directives may specify configuration settings, constants, changing fields, values and probabilities.

KStore data simulator directives and values processor process and files

The process of traversing a K may involve determining a match between a root node and a Result node of a node on the asCase list of a current K node. When learning is off and a match is not found, the procedure may ignore the particle being processed. An alternative solution determines which node on the asCase list is the most likely to be the next node. While the K Engine is traversing and events are being recorded into a K structure, a count field may be added to each K node to contain a record of how many times each K path has been traversed. The count field may be updated according to the processes traversing the K. Typically, the count is incremented only for learning functions. This count field may be used in determining which node may be the most (or least) probable.

Method for determining a most probable K location

The KStore is a datastore made up of a forest of interconnected, highly unconventional trees of one or more levels. One KStore can be added to an existing KStore to create an updated KStore by traversing the trees of the KStores and incrementing the count fields to reflect the added counts of analogous nodes. New KStore structure is added as needed. A KStore can be subtracted from an existing KStore to create an updated KStore by traversing trees KStores and decrementing the count fields in the updated KStore to reflect the subtracted counts of analogous nodes. KStore structure is removed as needed. Portions of a KStore can be added or subtracted from another KStore by isolating the paths to be used for updating. Instead of incrementing or decrementing counts, a dataset of individual records can be recreated from the KStore.

Adding and subtracting KStores

A scenario simulator processor receives a declarative file and invokes one or more data simulators to create one or more datastreams from a data structure may be built as specified by the declarative file. The declarative file may specify one or more scenario names, and a set of information corresponding to the one or more scenarios (one set for each scenario). Each set of scenario information includes one or more of the following pieces of information: parameters and settings for the data simulator and the number of threads to be started for each data simulator invoked.

Kstore scenario simulator processor and XML file

Typically, field names are saved separately from tables as metadata in modern databases. Databases did not traditionally get built into interlocking trees datastores that recorded the data as events. However, in cases where one may wish to do that, thus avoiding the need for saving separate metadata from the table data of the data base, a need was found to establish an identity for particular columns or fields when working with databases or sources of data that provide table data in field/record format. So, to build interlocking trees datastores from such records a mechanism to record such data was created, adding a column ID, preferably to each field within each record or sequence that is to be recorded. Putting the column ID or identifier is inserted into the record during particlization between each column variable. In preferred embodiments a delimiter was included between the column ID or field name and the field variable. Appropriate hardware and software systems were employed to implement the invention.

Facilitated use of column and field data for field record universe in a knowledge store

A tree-based datastore comprising a forest of interconnected trees is generated and/or accessed. The tree-based datastore comprises a first tree that depends from a first root node and may include a plurality of branches. Each of the branches of the first tree ends in a leaf node. Each leaf node may represent an end product, as described more fully below. A second root of the same tree-based datastore is linked to each leaf node representing an end product. Hence, the second root is essentially a root to an inverted order of the first tree, but the first tree is not duplicated. Finally, the tree-based datastore comprises a plurality of trees in which the root node of each of these trees is an elemental node, as described more fully below. The root node of each of these trees may be linked to one or more nodes in one or more branches of the first tree. The nodes of the tree-based datastore contain only pointers to other nodes in the tree-based datastore. The roots of the trees in the forest of trees comprising the tree-based datastore may be linked to data.

System and method for storing and accessing data in an interlocking trees datastore

A method for processing a particle stream in a KStore having a sensor level and a first level above the sensor level, including receiving a particle to provide a received particle and first determining a current K node to determine a partial sequence location. The method also includes second determining a match in accordance with the received particle and the current K node to provide a match determination and processing the current K node based upon the match determination. A new node is created if the second determining is negative. A current K node is set to the new node. A node count is incremented and a delimiter particle type of the received particle is determined. A match is determined between the delimiter and a Result node of an asCase node of the current K node to provide a delimiter match determination.

Method for processing sensor data within a particle stream by a KStore

An effective means for building an interlocking trees datastore from multiple data sources that contained within the datastore the knowledge of the origin of the data was needed, and is thus described herein. Should it be desired or necessary to identify the different sources of data within the Kstore data structure, the present invention teaches how an additional field for the origin of the data is created, how the origin of the data is stored in that newly created field, and how the interlocking trees data store takes that field and creates nodes within the structure, therefore embedding the origin information within the knowledge contained in the Kstore data structure.

Storing the repository origin of data inputs within a knowledge store

A method for optimizing the construction of a KStore includes providing a data set having a plurality of records with a plurality of fields, providing an optimization structure from a sample of the data set and analyzing the optimization structure to determine a field order for the records contained in the data set. A KStore is provided from the data set in accordance with the field order. The sample of the data set can include the entire data set, a random sample of the data set or a predetermined number of records from the data set. The data set has an initial order having earlier records and later records and the sample can include a plurality of the earlier records. The sample of the data set includes a user determined number of records from the data set. The optimization structure can be a KStore.

Optimizing the size of an interlocking tree datastore structure for KStore

A method for processing KStore sensors for use by a KEngine in a KStore system to process a K includes providing a stream of particles, instantiating a KStore sensor structure, identifying particles to be included in a sensor set and processing said identified particles. The KStore sensor structure can provide a correspondence between a particle and a sensor K node. Furthermore, the KStore sensor structure can be a list of K nodes, an indexed array or a hash table.

Method for processing a particle using a sensor structure

A method for accessing a KStore including at least one K path having at least one subcomponent node and at least one root node includes providing a utility routine and accessing the KStore using the utility routine. The accessing may include traversing the at least one K path. The at least one K path includes a further node and bidirectional links between the at least one subcomponent node and the further node. The method includes traversing between the at least one subcomponent node and the further node by way of the bidirectional links. The further node can be a beginning of thought node, an end of thought node, root node, or a subcomponent node. A location may be determined within the KStore. The K Store may be queried. Searches for information within the KStore may be performed. The KStore may be traversed to constrain and focus the KStore. The KStore may be updated.

Utilities for deriving values and information from within an interlocking trees data store

A set of mechanisms handles communication with a Knowledge Store and its K Engine(s). The Knowledge Store (Kstore) does not need indexes or tables to support it but instead is formed by the construction of interlocking trees of pointers in nodes of the interlocking trees. The K Engine builds and is used to query a KStore by using threads that use software objects together with a K Engine to learn particlized events, thus building the KStore, and these or other software objects can be used to make queries and get answers from the KStore, usually with the help of a K Engine. Under some circumstances, information can be obtained directly from the KStore, but is generally only available through the actions of the K Engine. The mechanisms provide communications pathways for users and applications software to build and/or query the KStore. Both these processes can proceed simultaneously, and in multiple instances. There can be a plurality of engines operating on a KStore essentially simultaneously. Additionally a mechanism for providing triggers allows for automatic reporting of events, conditions and occurrences to users and applications.

Functional operations for accessing and/or building interlocking trees datastores to enable their use with applications software

A method for transmitting information in a KStore system having a KStore, an application programming interface and an application layer wherein the information is transmitted between the KStore and the application layer includes transmitting the information between the KStore and the application layer by way of the application programming interface. The KStore is constrained by way of the application programming interface with at least one constraint to provide at least one selected K path. The KStore is focused to provide a further selected K path. At least one KStore parameter is determined in accordance with the selected K path to provide at least one determined K parameter. The constraining of the KStore includes traversing the at least one selected K path.

Growing community of inventors

Blue Bell, PA, United States of America

Jane Campbell Mazzagatti