|
The Active Concept Engine (ACE) is central to all the functionality in Ultra Knowledge Search Technology. It is a unique piece of proprietary technology that can automatically identify and prioritise the concepts it finds within unstructured text. The ACE offers a range of innovative search features that are built to assist site users find the content and people that interest them most.
When you use a light switch you don’t
think about how electricity is harnessed or how it gets delivered to your appliances, you just expect the light to come on. This analogy readily illustrates the thinking that has gone into the development of the functionality of the ACE. Requiring the user to learn arcane algebraic notation and a deep knowledge of the underlying engine is counter productive. Ultra Knowledge Search Technology is user-centric, designed to make search both simple and effective.
System Infrastructure
The ACE is written in C++
and can use peripheral languages and protocols such as XML, PHP, Ajax, C# and Flash when required to enhance communication and data transfer. Because of
this the ACE uses Windows Standard Server 2003.
The operational
limits of the ACE have not been challenged because it has been designed
to cope with very high levels of traffic and extremely large volumes of
data:
- Load balancing across multiple servers enables the ACE to respond properly under intense traffic conditions.
- A distributed architecture is available
to prevent deterioration of response times if large numbers of
documents are available for indexing. Rigorous testing has demonstrated
that the ACE functions well even when an index of 500 million
documents has been fetched and indexed.
ACE Components
The ACE is future-proofed by having a range of components that can be added as further functionality is required. Growth of a successful implementation is not limited.
Search Features
The user-centric approach to the design of the ACE allows Ultra Knowledge to offer a variety of ways for the user to interact with information. They range from simple non-disruptive interfaces that assist the user to find what they need, up to heavyweight research tools that give complete direction and control to the user while suggesting lateral or focused results. The modular structure of the ACE even enables Ultra Knowledge to create new custom interactions.
The diagram belows depicts some of the end user search features available from the ACE.
Core Technology
Since Ultra Knowledge was able to start with a clean
drawing board, considerable research was done to ensure that the best base
technology was used as the foundation of the ACE.
There were basically three choices for the core search technology:
- Database keyword search - where key words are pre-defined by the rigid structure of the content
- Semi-structured search - where a pre-defined taxonomy and ad hoc rules are constructed to establish the relationships between words
- Statistical search - where the relationships between words are defined by the inherent structure of the content that is indexed
Since the ACE is intended to work with unstructured content the first option was not viable.
The second option, variously know as
Semantic, Linguistic or Natural Language Processing, can give rise to
extremely accurate results but only when dealing with a static and well
pre-defined data set. Since the ACE works with totally undefined data that is
dynamic and changing by the hour, this was also not an option.
The Active Concept
Engine (ACE) is statistically based. This means that indexing and
maintenance are automated and it is not necessary to involve subject matter
experts to pre-define rules, dictionaries or word relationships.
With regard to the theoretical basis of the ACE, we have looked
at the work of Bayes and Shannon and we have also reviewed many papers
written by the dedicated academics who have researched statistical theory over
the past thirty years. However, the algorithms used in the ACE are not attributable to any particular theoretical direction, partly because they are highly influenced by the practical considerations imposed by working in the real world.
Digital DNA
The ACE is a sophisticated content
indexing and data storage engine that retrieves relevant content by comparing
"Concept Signatures" that are created for every indexed article. The concept signatures can be thought of as "Digital DNA" with
each article creating its own unique set of patterns. The engine uses
proprietary algorithms that enable the analysis and comparison of
content at multiple levels. A range of structures are created that reflect relationships at word, phrase, sentence, paragraph and
complete article levels.
The ACE is dynamic, and as new content is published the ACE automatically
updates the complex structures it uses, ensuring the user can always view up to date content.
|
