Custom Intelligent Discovery Solutions Powered By CCM™ and REGGAE
NdCore | NdCore|c Solutions (powered by CCM) | NdCore|r Solutions (powered by REGGAE)
ATS search solutions are based on our patented Contiguous Connection Model (CCM) and patent-pending REGGAE technology. Both CCM and REGGAE work with a wide variety of data sources, including relational databases. It uncovers relationships between data and provides mechanisms for investigating them. When loaded into a data warehouse, the resulting data views can:
- Find relationships implied by data
- Reveal relationships that were previously unknown, unobserved, or hidden
- Expose patterns in the data
Find the Missing Links with NdCore®
- NdCore is a scalable, extensible application and application framework powered by CCM and REGGAE.
- NdCore search technology lets you search for ideas, not words.
- NdCore excels in the integration of information from multiple disparate sources.
We Provide Scalable Solutions That Meet Your Specific Needs
Our custom applications tailor CCM and REGGAE technology to work seamlessly with your systems and data.
- Unstructured text search and analysis
- Automatic correlation of hierarchical data
- Intuitive data visualization tools
- Semantic data modeling
- An interactive, context-driven query wizard
- Text analysis/classification
- Search and "find similar" functionalities
- XML compatibility
Return to top
 NdCore - Searching for Ideas...Not Just Words
NdCore® is based on the premise that people want to search for ideas, not words. The focus is on integration of information from multiple, disparate sources. An example: When researching professional burglary, a search looking for articles containing the string “professional burglar” will miss too much relevant information, while a search for articles containing “professional burglar” or “professional” or “burglar” will drown the researcher in a flood of irrelevancy (while still missing meaty articles about “burglars”). What is needed is a discriminating tool that can find (and weigh) information relating to the concept of professional burglary: security alarms; fencing stolen goods; surgical gloves; recidivism … while excluding references to eye masks and black-and-white striped jerseys.
NdCore is that tool.
The Problem
Regardless of research topic, the problem is that there’s too much data to sift through. ATS provides two powerful tools to help overcome this problem:
- A high-level view that allows the user to browse through what is available without having a specific search goal in mind
- A targeted, directed search for information conforming to a particular data pattern, like in the above example
Before you can search for information, the data has to be in a searchable form: it has to have structure that software can use and understand. For NdCore, this means that the system needs to know how you think about your information, how you "model" it. Depending on the structure of your original information, its interrelationships, and the type of searching you want to do, you can choose from a number of existing data models or create your own. Once created or selected, the model tells the system how you think about your information. NdCore then uses this model to automatically discover all possible relationships existing in the information. It can discover relationships that were subtly or completely hidden, and permanently store them for quick, optimized retrieval.
Relational databases are sometimes capable of this type of analysis, but queries that describe induced relationships (relationships that were not defined in the original design of the data tables) have to be constantly recreated or rebuilt, resulting in massive processing overhead. With NdCore, all induced relationships are permanently stored, making for almost instantaneous information retrieval.
Once information is in NdCore, you can choose from a number of search techniques.
Click on the image for a larger view
For unstructured information (free form text) you can search by:
- Summarization - a high-level look at the information.
- Concept - specific ideas you are seeking. You can use this method to sift through your information with a "fine-toothed comb".
- Find Similar - once information of interest is found using either of the previous searches, you can then use the search result itself as a search-key to retrieve information about similar ideas.
For structured information you can search by:
- Trace - follow an informational element across documents over time.
- Discrete - discover where and how the individual elements within structured information relate or "converge," to discover hidden relationships in structured information
The Solution - Performance
If your requirements include high volume throughput in a real-time or near real-time environment, NdCore has a solution for you. The NdCore architecture can scale to meet the search techniques that best fit your needs. As your processing requirements increase, additional "Analysis Nodes" can be seamlessly added to your configuration to meet your demands. All NdCore network elements can be taken offline for maintenance without system disruption.
The Solution - Development
If you are developing solutions that are not addressed by NdCore but have need of our Referential Discovery technology, contact us via Email or 360-698-7100 and press "2."
Return to top
 NdCore|c Solutions - Powered By CCM™
Several years before XML became a de facto standard, ATS realized the need for a hierarchical, self-describing, data modeling and storage technology. ATS developed CCM to facilitate rapid information retrieval, enhance data navigation and visualization, and promote knowledge discovery.
CCM automatically correlates hierarchical data, modeling and storing it in Type:Value pairs (e.g. Actor:Kevin Bacon). A hierarchical database could have the following structure:
| Movie:American Graffiti |
|
Year:1973 |
|
Actor:Ron Howard |
|
|
Role:Steve Bolander |
|
Actor:Cindy Williams |
|
|
Role:Laurie Henderson |
|
Director:George Lucas |
CCM disassembles and stores this hierarchical structure as Type:Value pairs, but then goes much, much further: it creates a complete new hierarchical structure for each and every data point, with the data point as the root. These "inversions" are permanently stored and can be intuitively browsed using our powerful visualization tools.
CCM Concept Query
CCM end-users do not need to be accomplished SQL jockeys. Our interactive, context-driven query wizard allows easy, intuitive data browsing.
CCM Concept Query can analyze semantic usage and calculate semantic similarities between words. It can analyze a training set of documents to "learn the language" of a vertical community, and use what it has learned to augment searches.
The system can be trained from any coherent body of information, such as law, medicine, sports, finance, or marketing. Enterprises can also train the system based upon their unique language. Any language utilizing an ASCII character set can be analyzed. If you need to analyze unstructured text for concepts, or just need a simple query wizard, the CCM Concept Query module will very much simplify your job.
Return to top
 NdCore|r Solutions - Powered by REGGAE
REGGAE is a relationship generating graph analysis engine that gives data analysts an exciting new process to find context between data entities. Applications using REGGAE vastly reduce the amount of time required to obtain relevant information from your data sets.
Breaking the paradigm
Traditional databases excel at providing connections between nodes of data. Entities can be represented as rows and even whole tables of data that fully describe a specific entity.
But what about the context? How can you tell what relationships exist that define how data sets interrelate?
REGGAE accomplishes this by storing the contextual connections of data notes in a separate layer of information, called the Context Layer. With this context layer, REGGAE is able to quickly highlight interconnectivity between disparate sets of data.
Capabilities
REGGAE takes database analytics to a new level of capabilities in a single engine. Instead of focusing on the storage capabilities of the database, REGGAE is built to focus on the analytic capabilities. REGGAE is a fundamental advancement in relationship analytics technology in that it provides the associative search capabilities of a graph database, the tabular capabilities of an RDBMS, while also allowing the multidimensional analysis seen in a multidimensional database. It provides a solution specifically tailored and “normalized” for analytics.
Key features include:
- Advanced entity relationship analytics
- Two-tiered context-based graph architecture built on commercial relational database
- Automated collaborative analysis and enterprise environment
- Data mining (Graph, Similarity Analysis, etc.)
- OLAP capabilities
- True multidimensional analysis and dynamic modeling
- Relationships generation through entity aggregation
- Foundations on relational, graph, and multi-dimensional databases
- Easy integration with existing data stores
- Focused on end-user interaction with no SQL required
 As mentioned above, REGGAE is physically built on top of traditional RDBMS data structures, but essentially operates as a graph data structure. REGGAE uses a graph structure on two different tiers, the data layer and the context layer. (see sidebar)
Instead of graph edges connecting two entities together directly, graph edges always interconnect between the two tiers. This creates a graph data structure that then defines context for all entities. It effectively behaves not only like a graph database (for data mining), but also allows for capabilities commonly seen in relational and multi-dimensional databases as well. Each context node functions as a record pointer – such that a tabular structure may be extracted. Furthermore, because the links are explicitly stored, hierarchical report building is a simplistic traversal of the two graph layers (as seen in Multi-dimensional databases).
The context layer also serves as an additional index into overall database. This effectively allows you to analyze the connections of the data as well as the data itself. This layer also optimizes the performance of advanced data mining operations that would normally be very strenuous against a traditional RDBMS.
REGGAE also provides powerful collaboration capabilities by tracking all usage of the database and automatically identifying users with similar activity/patterns within the dataset. Any work a user has performed on the system is captured through these patterns and will identify intersections/overlap of analysis activity. Unlike other knowledge management and collaboration systems that require users to work in a specific manner or take an extra step to record their activity/observations in a separate knowledgebase, REGGAE delivers valuable collaboration information as a by-product of letting each user work in whatever manner best suits their needs, responsibilities and experience.
In summary, REGGAE is a highly capable data-mining engine optimized for aggregation of multiple data sources in a single view, user-friendly entity relationship analytics, and collaboration across multiple users, departments or organizations.
 Interoperability
Among the strongest features of the REGGAE engine is its ability to ingest sets of data contained in disparate data sources.
The REGGAE engine is capable of simultaneously incorporating data from a variety of sources, including:
- Any relational database (i.e. Oracle, SQL Server, MySQL)
- XML flat file databases
- Any other structured data source (Excel, comma or tab delimitated text files)
Use Cases
The REGGAE engine can be applied to a wide range of use cases. Each scenario listed below is indicative of the flexibility that connection discovery conducted through the REGGAE engine provides for data analysts.
Bioinformatics
Scientists have struggled with the vast amounts of data necessary to sort through in order to obtain relevant findings in the fields of genomics, sequence alignment, and medical history analysis and prognosis. By utilizing the REGGAE engine, scientists could quickly identify previously unknown connections between genome pairs, or recorded drug interactions.
Law Enforcement
Detectives record a huge amount of details about a case in the course of their investigations. Location, witnesses, weapons used, event time, etc. However, it’s not a simple task to correlate this information with other disparate sets of data about seemingly unrelated criminal activity. By applying the REGGAE engine to a law enforcement investigative system, detectives may discover new leads on cases based on exposed data they never would have associated with the criminal activity.
Terrorism Social Networking Analysis
Social networking is not simply the purview of MySpace, FaceBook or LinkedIn. The REGGAE engine could be applied to a system devoted to discovering the connections between members of known terrorist organizations. By identifying undiscovered relationships and connections between key members of terrorist networks, the data can be used to help government officials enhance security, and enact new measures to cripple terrorist threats.
Buyer recommendation systems
Understanding the likes and dislikes of a customer requires a multi-faceted method to manage the vast amount of information gleaned from normal consumer purchases. By supplying a tool utilizing the REGGAE engine with disparate customer data stores, a data analyst would be able to correlate a buyer's interest in music, media, browsing history, similar customer purchase histories, and other dissimilar variables, and produce information to create targeted advertising, movie recommendations, or other previously undiscovered customer relationships.
Each of these use cases illustrates the flexibility afforded by employing the REGGAE engine to power custom modules tailored to meet the needs of data analysts.
Return to top |
