Cubicon Platform
First Release Capability
Architecture Rational

Cubicon is a well-developed architecture based upon twenty years of applied research that will far exceed market expectations based on its ability to directly address a demanding set of key technological, business and social criteria.

The Cubicon Platform
The Need for a Cubicon Semantic Net Overlay
Cubicon Contextual Knowledge
User Knowledge Navigation
Cubicon Architecture for Learning at Web-Scale
Agent Executes within CubeRun Context Engine
Populating a Semantic Net through Automation
Intrinsic Revenue Generation Mechanism
Conclusion

The Cubicon Platform

Cubicon platform components:

— CubeRun —
Context engine provides interoperability to bind billions of devices

— CubeStudio —
Iconic IDE for complex systems engineering

— CubeNet —
Generative environment natively supports “cloud computing”

The Cubicon Platform is a radically simplified network architecture that enables true semantic functionality that can effectively manage the complexities of a universal medium to facilitate data, information and knowledge exchange.

Key notions of the platform are not restricted to any current Semantic Web standards but to emerging semantic technologies that can be applied on top of the current Web. The key notion of semantics is to hold knowledge representation separate from the content or behavior artifact. To create a representation that both people and machines can interpret.

Current research efforts towards ambient intelligence, next generation Internet and Web 3.0 / Semantic Web efforts are going in the right directions to support a semantic desktop, Wiki and mobile applications, but have a long road ahead. Ambient intelligence, NLP (Neuro-linguistic programming) based search, context-aware mobility, grid, mesh and remote sensor efforts are converging to address the need for knowledge representation that will be visible to and interpretable by other processes. The additional challenges of multi-core processing, deep packet inspection and security add to the complexity. The industry must move away from current rigid black box architectures like procedural code, objects and stack architectures that add up to a Tower of Babel. An amalgamation of technologies will be required to achieve the scale, performance, autonomic and sustainability properties necessary to handle the additional complexities of an the upcoming Net-centric Knowledge Age.

Traditional systems engineering has been exhausted. Textual expression mediums such as C, C++, Java and other languages are one dimensional and serial. Developer comprehension has been severely limited by sheer complexity of these system representations. Productivity suffers, especially in community development projects.

The Need for a Cubicon Semantic Net Overlay

Search engines access surface Web, while the potential for deep Web access is several orders of magnitude larger and not currently indexed using traditional search capabilities.

Search engines, such as Google, rely on automated crawlers and are efficient for finding Web pages based on keyword recognition. A However, a Web page search typically cannot reach information within a database. Its content is are only retrievable through the database's own search engine. Recognizing the distinction between searching Web page content and database content is important because the bulk of authoritative information remains inaccessible, buried in databases within the deep Web. Cubicon is designed to access distributed resources there, enabling a single query to launch navigations distributed spread across any number of databases. This would allow the gathering and retrieval of deep contextual knowledge to users impossible using traditional methods.

The Invisible Web figure depicts the Web as an iceberg. The visible tip is the surface Web, an estimated 10-12 billion pages currently indexed by various search engines. The deep Web contains more than 600 billion pages and lies below the surface, not accessible using any current public search technology.

The Cubicon Semantic Net collapses layers 4 through 7 of the ISO Reference Model protocol stack. Sitting on top of TCP/IP, it performs the transport, session, presentation and application functions. It interacts with the HTTP protocol and WWW W3C specifications to enable context to be applied to data and information content. If Cubicon overcomes the key technical challenges of representational complexity, the natural language problem, bottom-up assumption and the standards issue through fundamental computer science innovations. In-depth analysis of these challenges is are found in the companion Semantic Net Architecture paper.

The Semantic Net overlay is composed of topic maps that are distributed across Internet nodes ranging from mobile to router devices. Each topic map is controlled by a person or organization entity and represents knowledge detailing their domain of practice.

Cubicon Semantic Net overlay is composed of topic maps that are distributed across Internet nodes ranging from mobile to router devices. Each topic map is controlled by a person or organization entity and represents knowledge detailing their domain of practice.

Context emerges through topic associations. These links are navigated by 'intelligent' agents that understand a topic's meaning within an entity's context. Occurrence access can be gated and metered providing the a means for the entity to derive new form of license revenue for their content and services.

Cubicon Contextual Knowledge

Traditional OS directory systems organize files as nested hierarchies whereas knowledge concepts are organized into both taxonomies and graphs. A concept is an abstract, universal idea or notion that is represented in Cubicon as a 'subject'. Knowledge represented in its contextual form brings meaning to information. It creates new and powerful ways to organize concepts by their multiple aspects, allowing IP sharing that cuts across OS and organizational boundaries.

Context Resource Forms. Context resources will be distributed and linked as a grid across the current Internet to allow the formation of a Semantic Net. Cubicon Contextual Knowledge is represented in three fundamental resource forms:

Concept topology organizes subjects and their characteristics, ranging from general to specific taxonomy, controlled by an entity. Taxonomy captures the natural relationships between concepts representing inorganic, organic and abstract systems. An entity maintains its own context resources in their Community Repository. Multiple entities may co-develop and share taxonomy through community membership.

A topic map graph represents knowledge about an entity's domain of practice using topics (representing any subject), associations (representing the relationships between them) and occurrences (which denote relationships between topics and information resources relevant to a chosen subject). An association provides the meaning for a topic pair. Linked information sources are called content resources that can be based internally and externally to the entity's system. External content resources are controlled by other entities. Topic associations can take place between maps controlled by the same or different entities, creating a distributed grid for the emerging Semantic Net.

A situation service is a program that navigates between topics, drawing inferences about their linked content resources. The inference process arrives at some conclusion. Though it is not logically derivable from the assumed premise, it possesses some degree of probability relative to the analysis. An agent can be shared among community members and enables it to power custom search engine abilities that will navigate the peer-to-peer grid of topic maps. Agents provide open behaviors far beyond the limited proprietary algorithms hosted by Google and other current Web search services. Whereas search is the process of traversing keyword indexes in a server farm, navigation is the process of an autonomous agent traveling from topic-to-topic hosted by entity devices far surpassing the constraints of current indexed page ranking.

An agent is a representation of active knowledge that expresses itself by the manner in which it navigates through topic maps. This is in contrast to search that can only search sort through a static index of keywords.

An embedded agent appears as a 'smart' page link. Instead of calling and displaying a Web page, it activates a nested agent that might ultimately display just a single or many relevant Web pages after completing a deep inference on a set of content resources. Embedded agents are transparent to a Web browser user. This embedding will enhance the overall search experience by eliminating browsing steps. Over time, more and more Web pages will be accessed through the Semantic Net overlay, bypassing the need for indexed search services.

Agents are declared in the following iconic form and are completely syntactically-driven and semantically-bound. This means that a developer does not need to enter syntax and the language environment has the ambient intelligence prohibiting operational sequences that are outside of what the context engine knows how to execute. This environment keeps the developer on a set of invisible rails that do not constrain his/her ability to express what is in their mind. This level of program automation capability greatly expands the population of developers to include those with little or no programming experience.

Context Resources Summary. Agents interact with knowledge in the context space from multiple aspects. An aspect is represented by associated topics and their occurrence links to content resource. This grid overlay information liberates information from silos and walled gardens that is then transformed into knowledge that will be accessible and sharable at the concept level. This knowledge access is much deeper and meaningful than the Web we know today.

In the Web world, each device has an OS that provides walled access to its content through a hierarchical file system. Hence the current interest in semantic desktops.

Applications on these platforms provide access to the information they have knowledge of, but are not easily combined with others when such link ups were not planned in advance. Hence, the current interest in mash-ups.

It's very difficult in a world of walled gardens to reuse functionality (or logic) from one application to another, because it is locked into black box objects and compiled procedure code. This separation is fueling the current interest in Service-Oriented Architecture (SOA).

It is hard to imagine continuing to extend shared access, information and capabilities across the Net, especially in a current hostile atmosphere of constant change. This environment is fertile for change implementing the stability of future semantic technologies that will enable mobility, grid and the next generation Internet.

User Knowledge Navigation

At the Net scale mesh level, there will be countless ways to constrain, view and manipulate knowledge from any concept or set of concepts. Constraint browsers will become a necessary art form. Their creation will steadily occupy the next generation of Net developers. How people access, navigate and manipulate very dense knowledge space is the nexus for the upcoming technology race. Cubicon's advantage is that it provides the prerequisite context space where all concepts can reside and are associated.

Every conceivable useful UI is supportable directly and adaptively from the Cubicon context space. Cubicon is the generic model on which any conceivable user interface can be supported. With a constant stream of emerging new technologies targeting various markets (such as the current preoccupation with virtual reality) it is hard not to lose sight that many (if not all) will not scale to accommodate future Semantic Net needs. Cubicon merges programming, UI and knowledge development into a new art form. This iconic expression medium allows the emergence of significant new interface forms. It can cross map any kind of established knowledge with current UI capabilities to create 3D immersive spaces, reality browsing and quiet computing. These UI capabilities make complex real time information and media rich experiences both tractable and navigational to the user.

Interface Details. There remains a need for a generic interface into the context world without requiring the development of the requirement to develop a specific constraint browser, comparable to a current search experience.

Cubicon's contextual discovery interface will enable a user to enter a series of keywords mapped to subject terms that identify context resources available from foreign communities. This matching process can also take place on subject synonyms and in multiple natural languages and dialects. This discovery process is performed on the Context Registry operated by Cubicon.

Cubicon Architecture for Learning at Web-Scale

Cubicon provides a mechanism for human and machine-mediated system learning at Net-scale.

Cubicon provides a mechanism that is analogous to how people normally assimilate knowledge when they learn or share what they know. Learning is a process where concepts can emerge from personal insight or are brought to an individual's attention from an external source. Below we outline this questioning process:

Agents perform this learning process within the Semantic Net. Topic maps are living and morphing graph structures that form a mesh between entities that adapt and thrive in this rich environment. A resource-binding agent is bound to a topic. It can crawl external content, analyzing for specific patterns and link occurrences as content resources. A situation service traverses the maps and infers based on both topic associations as well as content resource analytic properties.

Learning on Micro and Macro Levels. The learning process occurs at both micro and macro levels.

The micro leaning process occurs when a situation service analyzes concept relationships. The agent can weigh the degree of association and taxonomy indirection as well as the analytic properties. Its conclusions can either be returned to the calling entity or it may be empowered to modify the topic map itself if necessary. This learning can be declared as induction, abduction and deduction behaviors.

The macro leaning process occurs across the Semantic Net and is harmonized by the Context Registry described below.

An entity can 'actualize' (regard or treat an abstraction as if it has concrete or material existence) a subject as a topic within a map. The same subject can be actualized into other entity topic maps. The subject can also be actualized into topic maps controlled by other community members. In this manner, a subject's actualization is 'shared' between maps.

A shared topic is actualized from a subject that is based on the same concept. Therefore, a situation service can infer about the subject in different maps by traversing these implicit associations between shared topics. (Washington State 'aspect': tourism, history, political representation, road, etc.). On the other hand, the user or a program declares an explicit association between singular topics.

A shared term is found in different subjects that are matched from multiple taxonomies. A situation service can infer meaning about a term in different subjects by traversing these pattern relationships that are maintained within the Context Repository. (Washington 'meaning': State, DC, President, University, etc.).

Note that diverse communities will likely declare the same concept as independent subjects. (Washington State 'Use case': Tourism Bureau, State Handbook & Guide, PEW Research Center, US Department Transportation, etc.). The Context Registry informs communities of likely concept redundancies. Overlapping concepts are graphically portrayed to an arbitrator who can elect to take one of two actions: 1) Convert their community concept into a shared subject where the origin remains within their taxonomy tree. 2) Adopt the foreign community concept through a refer subject that appears in their taxonomy base. Sharing maintains taxonomy relevance and local subject description, whereas referring gives up description control to another community.

It is also feasible that algorithms will be developed that analyze foreign subjects to identify intersecting concepts that may be possible candidates for consolidation between communities. This consolidation capability is fundamentally essential to achieving the convergence of global semantic knowledge.

An Agent Executes within a CubeRun Context Engine

A conventional Web crawler process is quite different from a situation service process executing in the Semantic Net. A Web crawler executes on information processors and pattern detection reasoners within a closed service provider's server farm. A situation service executes in a context engine on a full spectrum of pattern, linguistic and knowledge reasoners within distributed space on entity-controlled devices.

Topics are semantically value-linked. This enables a machine to automatically and very efficiently navigate maps. Topic navigation supplants human manual search based upon language-linked word search indexing. Cubicon's multi-dimensional space identification technique provides universal knowledge assimilation across countries and cultures by capturing information within contents and theory within concepts. Topics can link to occurrences of meaningful related particular information and universal theory resources. "The sum of information and theory is knowledge." (Richard Ballard)

An entity must download CubeRun (a runtime context engine much like a plug-in) into each device under their control. A situation service executing within a CubeRun analyzes both a topic's relationships with other topics as well as related occurrences of content resources. It first applies predicates and then gathers conclusions that its places in its suitcase. Based on this analysis, it will then eventually return to its home device. Once home, three possible actions exist:

Agent execution is performed by core operations in a partitioned secure sandbox and will not compromise the host device environment.

Figure Details. There are four topic forms; singular, shared, foreign and alien. A shared topic has already been discussed. A singular topic is not shared in any other topic map. Foreign and alien topics are proxies to topics that originate in either a map controlled by the same entity or a different entity.

Aside from an association between a pair of topics, this pair relationship can be further defined with roles on either or both end points.

An occurrence links to a content resource in two ways: A Web page is addressed through its URI name, while a directory file is addressed through its file path name. A remote device storing a directory file must execute a CubeRun that is responsible for maintaining the occurrence link.

Populating a Semantic Net through Automation

This section explains the way that Cubicon Semantic Net converts Web and other informational sources into contextual knowledge. For a clean slate, there needs to be both user- and machine-driven methods for populating a Semantic Net with context resources.

To summarize, the Cubicon Semantic Net is composed of three context resource forms: concept topology (representation), topic map (relationship), and situation service (behavior).

User creation. CubeStudio is an immersive iconic environment for collaborative context-based systems development. CubeStudio users will be knowledge workers who understand their domain of practice and will require little or no previous programming background. Users create context resources collaboratively using Cubicon's iconic language (also referred to as an expression medium). Several hours of orientation and experimentation will provide users with meaningful results in developing concept topology and topic map components. Situation services require a higher level of proficiency and planning as an advanced level of semantic reasoning is involved.

Our first product release will be a Semantic Desktop that will be the driving force behind viral adoption of the Cubicon Platform based on current computing and knowledge management trends towards a Semantic Net. This social semantic interface will enable people and communities to directly collaborate with their peers while dramatically reducing the amount of time spent filtering and filing information in their personal knowledge space. A Semantic Desktop provides rich features for organizing, interrelating, integrating and sharing all personal information from local and remote content resource sources. The technology's differential capability is the ability to share ontology and automatically consolidate concept representations automatically across the globe. Only Cubicon has a credible architecture for achieving this global capability. The following list represents examples supported by the first product:

Current desktops have an underlying operating system based on a hierarchical model that provides a category tree to organize application behavior and structure files. Traversing a tree then provides file access through a nested directory index. An OS lacks the ability to organize files according to a graph model that categorizes by association. This organizational capability is required to support the manner in which humans naturally organize knowledge.

Virtualization techniques overcome many of the 'walled garden' constraints between OS types, but do not overcome the barrier of information flow between entity domains. This function is the purpose of Service Oriented Architecture (SOA). Unfortunately, its adoption has been retarded due to the lack of a true Semantic Net infrastructure that wouldrequired to support community development of ontologically-based services.

The Cubicon Semantic Net provides a graph model that enables entity domains to maintain semantic relationships via topic associations. Topic maps will overlay internally the application files as well as external content resources. This Semantic Net Environment (SNE) allows an entity to control one or more devices as physical nodes that each have has an IP address. An entity can maintain multiple domains and assign a set of nodes to operate under one of six possible topologies: discrete, proxy, assembly, server blade, provision or swarm. This topology demarcation is vital to provide the ability to perform net-centric processing.

Each device node will contains a CubeRun context engine that executes under an a traditional operating system. As more and more functionality is developed into the Semantic Net, more nodes will natively process all applications under CubeRun.

Machine creation. Existing Web page, source library, data schema, documentation, comment, metadata and other 'as-is' information assets may be analyzed through inference to identify knowledge tokens that are automatically turned into components. Developing context resources is a bottom-up process and provides the foundation to link topic maps to enterprise architectures and other governing models. Semantic discovery enables concept-level search using business terms, without having to know the (often cryptic) 'as-built' naming conventions established by legacy database programmers. This semantic-driven process combines domain modeling, linguistic ontology reasoning and other developing knowledge creation techniques. These semantically enabled behaviors provide a basis for recognition of topic map navigation opportunities that can then be exploited using situation services.

Whereas, a situation service reasons between topics, a resource binding agent is a behavior within a topic that examines termspace, association/roles and content resources for the purpose of automatically extending the occurrence of additional resources. This extension usually first requires the classification of the resource's content by matching document text with a predicate pattern. A match that fits the heuristics will generate a new analytic facet that links the annotated document passage or image back to the topic. A resource binding method can be shared with other community members and is easily adapted for use in other topics. We can create bindings with Google's Enterprise Developer APIs that will leverage the search engine to perform initial content resource qualification for the resource binding agent.

Cubicon will publish the complete API for both the Community Repository and the Entity Repository enabling an external program to create context resources through this mechanism.

Intrinsic Revenue Generation Mechanism

Cubicon Corporation anticipates that the first version of the platform will be distributed freely in order to gain early widespread adoption. However, the platform will contain multiple built-in mechanisms that enable Cubicon to be harnessed to generate revenues. One of the first incentives will be the ability to enable allow entities to charge for access to their high value topic maps. Cubicon will then retain a percentage of this service revenue for operating all exchanges. These mechanisms provide a gateway into a new a limitless new revenue-generating paradigm that will far exceed current advertisement placement supporting computing services.

Cubicon Platform First Release Capability Architecture Rational