Site News  |   Monitor Keywords  |   Monitor Archive  |   Organizer  |   Account Info  |

Assissted knowledge discovery and publication system and method

Assissted knowledge discovery and publication system and method description/claims

The present application claims priority to Canadian patent application No. CA 2,595,541, filed on Jul. 6, 2007 entitled “Assisted knowledge discovery and publication system and method” by the same applicant.

This invention generally relates to knowledge discovery, content creation, and content sharing using people, computer systems, software program agents, and databases.

Internet has provided a long awaiting tool for connection and communication of people around the world. One of the most important applications and implication of Internet is its use in enhancing ideas and rapid information exchange between people or groups of people with similar interests. Such growing interest has created many applications and systems for group discussions and question answering, such as Yahoo ask, wikipeida, search engines, photo and video sharing, numerous portals, discussion groups, and the like. These systems and applications have accelerated knowledge discovery, creation of artistic contents, producing novel and useful inventions, and in general advancement of our understanding of the universe around us.

However, since most of these knowledge sharing and contributions are arbitrary qualified, it takes time for general public to come to a robust and lasting understanding of a subject, or appreciation of a content. Therefore, the vast amount of data, that is being generated daily, has to be filtered out over a relatively long period of time by collective wisdom of public before it can be used. While in most subject matters of general public interest, ordinary people may contribute to the subject and let the fact and best solution to be found overtime, these unsupervised method of general public understanding growth lacks the rigor and credibility that is needed for a real advancement of public well being. The rigor and credibility only comes after a relatively long period of time. Mostly the information available through Internet needs further verification and research by the consumer and this could be time consuming and frustrating.

The process of peer reviewed scientific contribution publication, on the other hand, has the rigor and substance and therefore the credibility that is needed for true advancement of human knowledge, nevertheless it is a very slow process and does not present the speed and ease of accessibility that is necessary to tap into the vast potential of general public brain power and knowledge. Editors and reviewers of scientific journals do not have much incentive to serve unknown ordinary contributors. Moreover, naturally, they do not have the resources or expertise to find and cover all the subject matters of importance and assess and investigate all submitted contents.

Therefore there is a need in the art to have a system that, automatically or semi-automatically, can assist both publication/broadcasting administration and contributors to screen and assess all submitted contents in terms of their intrinsic value and substance before being viewed or used by public, without posing the above mentioned constraints. It is also desirable to have a system that can systematically guide users, through their research to discover, innovate, create, and make valuable contributions. It is also advantageous to have a central system that allow all the qualified experts launch their own publication/broadcasting ventures with the least amount of investment and overhead for commercial gain thereby accelerating the rate of knowledge discovery, knowledge distillation, and economic growth.

In this application a system and method is presented for knowledge sharing and discovery by analyzing the content of online repositories, building an association database of ontological subjects, and solicitation of electronic contents in the form of a text, audio, or video and any combination of them. The system and method can assist and guide the users and creators, regardless of their level of knowledge, to being able to make valuable contributions, while shortening the research and creation time significantly. The shared knowledge is peer reviewed by authorities in each subject so that their quality and substance is more reliable than arbitrary qualified contents presently available in the Internet.

The system is comprised of information processing units in the form of hardware and software that are connected to the Internet by communication means. The processing units can be comprised of electronic hardware such as CPUs (central processing units) memories, and software in the form of specialized programs and algorithms, and intelligent agent program, in any applicable computer language.

In building the system software agents are used to find important subject matters/fields of interest by looking up into a list of subjects gathered from various sources such as lexicons, ontologies, dictionaries, special dictionaries, and searching through Internet and counting and ranking the importance of a subject by counting the number of documents containing that subject or any other ranking methods for concepts. At the same time the software agent is looking for proper names and affiliations and addresses that are associated with the subject and ranking them accordingly based on their level of authority. Alternatively the system finds the subjects of importance and interests and the associated experts by directly searching through readily available databases where it can find the desired information such as university URLS, specialized professional associations, who's who, and all online publication collections available.

The system then assigns appropriate names or titles for such subject matters and makes a list of available subjects and titles as candidate name for publication/broadcasting shop to be used for subscription and running by users. In the preferred embodiment, the system further provides an online publishing/broadcasting format/s for each subject matter in the form of online journals or knowledge sharing groups, interactive conferences, broadcasting templates and the like, which is called a publishing/broadcasting shop in this application. The system further contains a database of authorities' experts in each subject matter for consultation and reviewing.

Users, who want to establish their own online publishing/broadcasting shop, then may apply to subscribe or buy online publishing/broadcasting shop's title/s among the topics and titles available. Alternatively the system accepts suggestion from interested users or subscriber to open a shop with their own suggested title or name. Interested users can include individuals, legal entities, a group of individuals as well as computer agents. The system will grant the privilege of establishing an online publication/broadcasting shop according to the system's predetermined standards. Once the application is approved and a title of publishing/broadcasting shop is assigned to the user the owner of the online shop can use the service of the system and start soliciting and providing the service to her/his group of people interested in that subject matter.

To assist the editors and contributors, reviewers, and users, the system has a distilled universal repository of human knowledge that is called Ontological Subject Map (OSM) in this application. The OSM is used to screen, evaluate, guide and assist, and measure the value of a submitted content, its novelty, and overall merit of a contribution. By consulting the OSM the system can pose useful questions and make intelligent suggestions and guides for further research or clarification.

The OSM is a layered indexed repository of universal knowledge that is built by indexing all related existing concepts and subjects, nouns, proper nouns, compound nouns, named entities or in general all such conceivable entities and concepts, that we call Ontological Subjects (OS) in this invention. The layered index or database is built by starting from one or a number of most popular ontological subjects and searching the available databases to find all other ontological subjects associated with each of them ordered by their association ranks (e.g. counts.) Then each ontological subject is indexed with a desired number of other ontological subjects in each layer ordered by their association ranking. Once this layer is constructed and indexed we repeat the procedure to find the most related OSs with each member of this layer. A node in an open 2-dimensional tree like graph may represent each OS. Each node therefore can only be connected to its above OS node and a number of other nodes below it. In each layer there are two types of nodes, namely Dormant or Non-Dormant (growing). In each layer a node is dormant if the corresponding OS is already been growing in upper layer/s or the same layer. In a situation and according to one exemplary embodiment, if more than one OS is found associated with several upper nodes, and it is not growing in an upper layer, then it will become Non-Dormant only under a single node which has the highest ranking association to its immediately above node. In this manner each ontological subject is growing only once in the whole index. Therefore each non-dormant node is connected to one node above and is connected to a number of nodes below it. Dormant nodes are only connected to its immediately above node. If desired number of associated OS was not found for a node, then we add extra nodes and mark them as unknown. The desired number of associated OS for each node can be arbitrarily selected. However, for simplicity we may choose a constant number of associations for each node.

Furthermore we may consequently represent an OS with a discrete spectral like function whose horizontal axis is the associated OSs and the vertical axis is the value of each associate. In this way an Association Value (AV) function is defined and stored in the database for each OS for further usage. The association value (AV) function can be considered a signature spectrum of an OS. Using signal-processing techniques, such as cross correlation, autocorrelation, Fourier Transformation (FT), Discreet Fourier Transformation (DFT) one then is able to extract the information and find a hidden relationship between OSs. For instance, using the concept of power spectral density, one may define and measure the power of an OS as a sign of its importance or for approximate reasoning application etc.

At the same time or after the indexing of OS association is completed, another software agent will look for the kind of associations between each OS and it's associates by searching through databases such as WordNet, FrameNet, the whole internet, or any such a database that a relation between an OS and its components is expressed by natural languages. The agent will look for patterns of explicitly expressed statements or semantic frames, as defined by FrameNet project in Berkeley University, to establish the kind of relationship between each two OSs. The agent may also use natural language processing (NLP) methods and algorithms such as text simplification, to find such an association pattern. However since there is a vast amount of data available, the chances are that the agent will be able to find the explicitly expressed and verified statement or frame, which is composed by humans, that is looking for. The verification of relations is done by statistical analysis of the database. Diversity of sources and a number of times that a statement is repeated to express a relation between two OS leads to the verification of that statement. These statements, or semantic frames, expressing a relationship between an OS and its components are also stored and indexed for further reference.

This database is then used to assess textual documents or any electronic content, such as audio or video, pictures, graphs, curves etc., that its information is transferred to textual format. The system first extracts the ontological subjects of a document and forms an OS spectrum or associated set for the document, with predetermined weighting coefficients rules. In one simple aspect of the invention, the system then can select an OS as the principal OS of the document and compare the document spectrum with that of the principal OS spectrum stored in the database, for further analysis. Alternatively one may partition a document to a number of parts and repeat the process of OS mapping to these collections of smaller content in the same way that an OSM is made from larger collection of contents.

The analysis includes, but not limited to, discovery of new ontological subjects, and discovery and verification of new associations between OSs. Over the time, new nodes and associations will show their importance by leading to growth of its newly discovered node or other nods, and finding the verified associations that are valuable to other contributors or is of commercial interest to commercial entities and ventures.

The system may also expand each OS to its constituent OS components and forms a more expanded OS spectrum for the document. In this way for each document we can form an almost distinguishable OS spectrum. The document OS spectrum bears important information about the value of the text compositions, its novelty and main points. Peaks and valleys may be used to analyze the content in terms of its novelty and an indication of possible new knowledge. For instance from the document spectrum we may select the highest amplitude OS as the main or principal subject of the text, then look at the next number of highest amplitudes OSs and form an abbreviated or abstracted spectrum of the text. Then compare this abstracted spectrum with the spectrum of the main OS already stored in the database, if there is a strong correlation between the abbreviated spectrum of the text and the principal OS spectrum in the data base, chances are that the content of the text does not bear much information. However for further checking one may look at the kind of statement and frames that is been used in the text to connect the components of the document spectrum to the main OS and compare it with the existing database of known relations between the these OSs. Generally there are more ways known in the art of spectral and signal analysis to evaluate the correctness and novelty of the text using the mentioned OS spectrum. When there are distinguishable peaks in the document spectrum that system does not have a record of verified relations for them, then the system mark them as novel and worthy of investigation and can compose a series of questions or suggestion to explain their relationship. It may also zoom to less amplified OSs and question and suggest a relationship between a high amplitude OS with a lower ones etc. All these information are available both to the editors of each shop and the creator of the content. The system or the editor of each shop can present such unknown to the public and solicit for contributions to the solution.

The strength of such a knowledge discovery system lies in its systematic processes, large number of potential participants, limitless subject matters, and its vast databases that are not readily available to individuals. The potential value of the system also lies in that the method enables measuring and quantification of one's contribution, both implicitly and explicitly to the advancement of the knowledge database.

• Provisional Patent
• Utility Patent

• What Is a Patent?
• What Is a Trademark or Servicemark?
• What Is a Copyright?
• Patent Laws