63 results about "Factoid" patented technology

An evidence evaluation method and system based on question answering converts a report of analyzed information and / or a model of information into a collection of questions, determines answers for the collection of questions. A fact in the report is marked as being supported if one or more of the answers for the collection of questions support the fact. A fact in the report of analyzed facts is marked as being refuted if one or more of the answers for the collection of questions refute the fact. The method and system also may collect the answers as evidence and add the evidence to the model of information to create an updated model of information. The steps may be repeated using the updated report and updated model.

A fact extraction tool set (“FEX”) finds and extracts targeted pieces of information from text using linguistic and pattern matching technologies, and in particular, text annotation and fact extraction. Text annotation tools break a text, such as a document, into its base tokens and annotate those tokens or patterns of tokens with orthographic, syntactic, semantic, pragmatic and other attributes. A user-defined “Annotation Configuration” controls which annotation tools are used in a given application. XML is used as the basis for representing the annotated text. A tag uncrossing tool resolves conflicting (crossed) annotation boundaries in an annotated text to produce well-formed XML from the results of the individual annotators. The fact extraction tool is a pattern matching language which is used to write scripts that find and match patterns of attributes that correspond to targeted pieces of information in the text, and extract that information.

A query directed password scheme is disclosed that employs attack-resistant questions having answers that generally cannot be correlated with the user using online searching techniques, such as user opinions, trivial facts, or indirect facts. During an enrollment phase, the user is presented with a pool of questions from which the user must select a subset of such questions to answer. Information extraction techniques optionally ensure that the selected questions and answers cannot be correlated with the user. A security weight can optionally be assigned to each selected question. The selected questions should optionally meet predefined criteria for topic distribution. During a verification phase, the user is challenged with a random subset of the questions that the user has previously answered and answers these questions until a level of security for a given application is exceeded as measured by the number of correct questions out of the number of questions asked. Security may be further improved by combining the query directed password protocol with one or more additional factors such as Caller ID that assure that the questions are likely asked only to the registered user.

A DeepQA engine is enhanced to provide a digital medical investigation tool which assists a medical professional in researching potential causes of a set of patient conditions, including clues, facts and factoids about the patient. The DeepQA engine provides one or more answers to a natural language question with confidence levels for each answer. If a confidence level falls below a threshold, the enhanced DeepQA engine performs a crowd sourcing operation to gather additional information from one or more domain experts. The domain expert responses are provided to the medical professional, and are learned by the enhanced DeepQA system to provide for better research of similar patient conditions in future queries.

In order to provide a non-factoid question answering system with improved precision, the question answering system (160) includes: a candidate retrieving unit (222), responsive to a question, extracting answer candidates from a corpus storage (178); a feature vector generating unit (232) for generating features from combinations of a question with each of the answer candidates; SVMs (176) trained to calculate a score of how correct a combination of the question with an answer candidate is, upon receiving the feature vector therefor; and an answer ranker unit (234) outputting the answer candidate with the highest calculated score as the answer. The features are generated on the basis of the results of morphological analysis and parsing of the question, a phrase in the question evaluated as being positive or negative as well as its polarity, and the semantic classes of nouns in the features.

A system automatically generates OLAP metadata objects for a multidimensional analysis program such as OLAP from SQL statements without involving the database administrator, and eliminates the need for database administrators to manually perform this analysis and design. The system deconstructs a set of SQL statements into tables and aggregates metrics for measures and joins. It recognizes that the relational database structure contains both dimension tables and fact tables. Over many SQL statements, fact tables will have a large measure metric while dimension tables will have a low measure metric. Tables are linked based on large join metrics; small join metrics are ignored. The present system builds the OLAP cube model from the fact tables, dimension tables, and joins. The analysis of the SQL statements allows the system to map the hierarchies within the dimension table, and provides the system with information about attribute relationships and cubes.

A virtual interview is conducted in the context of a legal matter, for example, using online forms, thus enabling rapid and large-scale information capture from distributed recipients. A user can define questions and virtual interviews once, and use them again and again; can define an approved set of questions to insure consistency of interview practice; and can de facto improve the consistency and credibility of information gathering in legal matters. Users can pre-establish action items that are automatically created upon receiving specific responses to virtual interviews. Interview response may be managed by exception, i.e. exception alerting, and tools are provided to view exceptional cases for follow-up action. For example, alerts can be issued on the basis of such factors as: if any one responded that they had data at home, send an alert to the attorney and an action item to do an appropriate collection; and if anyone responds that they know others in scope, initiate an action item assigned to paralegal to add identified individuals to scope. A user can conduct many interviews systematically to determine where all key facts related to the matter are located; to update information in a highly distributed way for effective legal management; and to manage responses efficiently.

OCR errors are identified and corrected through learning. An error probability estimator is trained using ground truths to learn error probability estimation. Multiple OCR engines process a text image, and convert it into texts. The error probability estimator compares the outcomes of the multiple OCR engines for mismatches, and determines an error probability for each of the mismatches. If the error probability of a mismatch exceeds an error probability threshold, a suspect is generated and grouped together with similar suspects in a cluster. A question for the cluster is generated and rendered to a human operator for answering. The answer from the human operator is then applied to all suspects in the cluster to correct OCR errors in the resulting text. The answer is also used to further train the error probability estimator.

A DeepQA engine is enhanced to provide a digital medical investigation tool which assists a medical professional in researching potential causes of a set of patient conditions, including clues, facts and factoids about the patient. The DeepQA engine provides one or more answers to a natural language question with confidence levels for each answer. If a confidence level falls below a threshold, the enhanced DeepQA engine performs a crowd sourcing operation to gather additional information from one or more domain experts. The domain expert responses are provided to the medical professional, and are learned by the enhanced DeepQA system to provide for better research of similar patient conditions in future queries.

The invention relates to improved solutions for information retrieval, wherein the information is represented by digitized text data. This data is further presumed to be organized in terms (431-438), documents and document corpora, where each document contains at least one term (431-438) and each document corpus contains at least one document. Based on a concept vector (420-424), which conceptually classifies the contents of each document, a term-to-concept vector is generated for each term (431-438) in the document corpus. The term-to-concept vector describes a relationship between the term (431) and each of the concept vectors (420-424). On basis of the term-to-concept vectors for the document corpus, a term-term matrix is generated which describes a term-to-term relationship between all the terms (431-438) in the document corpus. The term-term matrix may then be processed and used for retrieving information from the document corpus, such as the fact that a first term (431) is related to a second term (436).

[Object] To provide a question-answering system capable of appropriately answering to a non-factoid question.[Solution]A question-answering system 30 includes: a related document searching unit 54 responsive to a question, for taking out answer candidates from an object document storage unit 32; an answer candidate extracting unit 56 extracting plausible ones from the answer candidates; a causal relation recognizing unit 60 recognizing causal relation expression included in extracted answer candidates; causal relation relevancy determining unit 64 determining whether or not a recognized causal relation is relevant as an answer; a feature vector generating unit 66 generating a feature vector related to a combination of the question and the searched answer candidate using the results of processing up to the causal relation relevancy determining unit 64; and a feature vector generating unit 66 for calculating, when a feature vector is given, a score indicating a degree of how correct an answer candidate is as an answer to the question, for the question and the answer candidate as a source of the feature vector, and for outputting, as the answer to the question, an answer candidate having the highest score.

The present invention provides attorney terminals which operate using an outline for storing, associating and managing case evidence, case law and work product for a given lawsuit at issue. Accessed through attorney terminals, the outline is structured based on a hierarchical categorization of the lawsuit into the law and fact at issue. Associated with each categorization entry in the hierarchical outline are groupings of case law, case evidence, relevance and draft discovery information for rapid access by the attorney. Each categorization entry in the tailored outline provides instant access to case law via headnotes, treatise selections, seminal cases, and preset searches. The disclosed invention also automatically: 1) tracks the use of Exhibits in a proceeding; 2) generates draft portions of a pretrial order including jury instructions; and 3) generates time-lines for analysis and use during a proceeding. Draft interrogatories, document requests and deposition or trial questions are also provided.

A medical image storage and retrieval system includes a database with relationally linked tables including a disease factoid table, an image and image caption table, and a patient data table. A flexible system allows for peer review, remote access and maintenance of the stored data, and query searching and retrieval of groups of related multimedia (image and text) case file information. The system facilitates distance learning and remote consultation.

A game module 12 is added to a home page 11. The game results are fixed with the prize winners on a real time basis. When a game is over, the customer knows immediately whether the customer has won a prize. If the customer has won a prize, the prize is given to the customer on the spot and the customer's name is added to the prize winners' list. Repetitive challenges may increase advertising effect of the home page because the home page would be looked at repeatedly. If the facts of winning the prize are displayed on the applicants' terminals on a real time basis, the applicants would rely on the game. The customers who have won the upper prizes are apt to let their acquaintances know their winning, which would enhance the number of customers who look at the home page.

Systems and methods are presented that may involve receiving a causal fact dataset including facts relating to items perceived to cause actions, wherein the causal fact dataset includes a data attribute that is associated with a causal fact datum. It may also involve pre-aggregating a plurality of the combinations of a plurality of causal fact data and associated data attributes in a causal bitmap. It may also involve selecting a subset of the pre-aggregated combinations based on suitability of a combination for the analytic purpose. It may also involve storing the subset of pre-aggregated combinations to facilitate querying of the subset.

A medical image storage and retrieval system includes a database with relationally linked tables including a disease factoid table, an image and image caption table, and a patient data table. A flexible system allows for peer review, remote access and maintenance of the stored data, and query searching and retrieval of groups of related multimedia (image and text) case file information. The system facilitates distance learning and remote consultation.

A reality-based hypothetical is developed and provided to a set of training participants. The reality-based hypothetical is a fact pattern describing an original lawsuit involving a set of real life scientific studies, data, witnesses, and issues that have never been litigated where only the claims, plaintiff, and defendant are fictitious is a training device for teaching a set of target legal and science objectives, such as determining the scientific integrity of proposed scientific evidence and understanding the scientific evidence as a whole. The “reality-based hypothetical” is developed to provide practice in performing problem solving techniques to determine the “scientific integrity” of proposed scientific evidence and scientific testimony. One or more dialogue sessions are provided to gather information necessary for making a determination on the admissibility of the scientific study. A decision on the admissibility of the scientific study can be rendered during each dialogue session, which occurs at designated segments during the training.

The invention discloses a construction method of a question-answering system based on document set multi-hop reasoning. The invention belongs to the technical field of natural language processing. Themethod comprises a training stage and an operation stage; in the training stage, question and answer pairs and support facts which need to take a plurality of scattered texts in a plurality of documents as support evidences are collected to form a training set, related support entities and answers are found from a given document, and a model is optimized by using a back propagation algorithm after comparison; and in the operation stage, corresponding answer prediction values and support facts are fed back to the user through the client and the server. In the construction method, the finally collected data is presented in the form of natural language based on documents instead of knowledge base questions and answers, the content and difficulty are diversified, the training time of the model can be effectively saved, dynamic reasoning is carried out on the GNN graph network, the sequence information of the text is well extracted, and the questioning and answering effect is better.

An automated system updates electronic medical records (EMRs) based on dictated reports, without requiring manual data entry into on-screen forms. A dictated report is transcribed by an automatic speech recognizer, and facts are extracted from the report and stored in encoded form. Information from a patient's report is also stored in encoded form. The resulting encoded information from the report and EMR are reconciled with each other, and changes to be made to the EMR are identified based on the reconciliation. The identified changes are made to the EMR automatically, without requiring manual data entry into the EMR.

The present invention provides a dimension table processing apparatus that efficiently derives a dimension hierarchy for a denormalized dimension table. The first information presenting means presents information on a plurality of fact tables included in a multidimensional database to an administrator as first information. The second information presenting means receives a fact table selected by the administrator from among the fact tables related to the first information, and presents information on dimension tables associated with the selected fact table in a star schema to the administrator as second information. The dimension hierarchy deriving means receives a dimension table selected by the administrator from among the dimension tables related to the second information as the one to be processed, calculates a predetermined evaluation value for each column in the dimension table to be processed based on data in each cell belonging to that column, and derives a dimension hierarchy of the columns, which consists of a plurality of hierarchical levels of the respective columns, based on the evaluation values thereof. The third information presenting means presents the dimension hierarchy for the dimension table to be processed derived by the dimension hierarchy deriving means to the administrator as third information. The approval receiving means receives a dimension hierarchy approved by the administrator based on the third information as an approved dimension hierarchy.

In an embodiment of the present invention, a GraphSLAM-like algorithm for signal strength SLAM is presented. This algorithm as an embodiment of the present invention shares many of the benefits of Gaussian processes yet is viable for a broader range of environments since it makes no signature uniqueness assumptions. It is also more tractable to larger map sizes, requiring O(N2) operations per iteration. In the present disclosure, an algorithm according to an embodiment of the present invention is compared to a laser-SLAM ground truth, showing that it produces excellent results in practice.

A message processing system 1 processes messages 5 such as emails being delivered across a network. A plurality of processing modules 10 are each operable to perform an action. A policy engine 11 causes the operation of processing modules 10 selectively in accordance with rules in a rules data store 12 and facts in the fact data store 13. The rules specifying the performance of actions in dependence on facts. The actions performed by the modules 10 include actions of analysing a message 5 and generating message facts specifying information about messages 5, such as the presence of unacceptable content. Thus the rules may specify actions dependant on such message facts. The actions include actions of controlling the delivery of a message 5 or other remedial action.

Queries describe users' search needs and therefore they play a role in the context of learning to rank for information retrieval and Web search. However, most existing approaches for learning to rank do not explicitly take into consideration the fact that queries vary significantly along several dimensions and require different objectives for the ranking models. The technique described herein incorporates query difference into learning to rank by introducing query-dependent loss functions. Specifically, the technique employs query categorization to represent query differences and employs specific query-dependent loss functions based on such kind of query differences. The technique employs two learning methods. One learns ranking functions with pre-defined query difference, while the other one learns both of them simultaneously.

A computer-based assessment that presents basic math facts in an operation and records the amount of time taken to answer each fact correctly. By measuring the latency of the response, the program can accurately determine the facts that are being recalled from memory and those that are solved using a counting or other procedural strategy. Once an initial assessment is completed, a grid is constructed that allows the student and teacher to see the fluent facts as well as those facts that were answered slowly and / or incorrectly. The system uses the grid to begin instruction on the non-fluent facts. Math facts are systematically presented, thereby instructing the student until the facts are mastered.

A system and method permitting users to author, publish, and render a dynamic interactive web page. The dynamic interactive web page is generated by a knowledge object which is a block of information composed of a plurality of action-condition rules. Knowledge objects may be used to build, but are not limited to building, question and answer facts, multiple choice trivia's, multiple choice quizzes, multiple choice test, polls, surveys, consultations, conversations, or complex problem resolution systems through series of information exchanges between a user and a knowledge object. Knowledge objects may provide needed knowledge in, but are not limited to, the following categories: Education, physics, history, geography, sciences, mathematics, mechanics, electronics, business, finances, entertainment, health, hobby, games, parenting, relationships, religion, real estate, shopping, sports, styles, toys, travel, trivia, etc. . . .

The invention provides a remote automatic correcting method for circuit virtual experiments, which comprises the following steps of: resolving input standard answer experiment scripts to obtain a set of apparatus score points and a set of circuit connection score points; selecting the required score points from the score point sets, and setting a score corresponding to each score point, wherein after the setting is finished, the score point is a score rule and a plurality of score rules form a score rule base; resolving virtual experiment scripts to be corrected into apparatus objects and apparatus port objects, and converting the apparatus objects and the apparatus port objects into factual data; judging whether the factual data are matched with the score rules in the score rule base, and recording the matching result of each score rule during matching; and accumulating the scores of the successfully matched score rules to obtain a total score of experiment correction, and outputting the matching results and the scores serving as corrected score sheets.

The present invention provides a dimension table processing apparatus that efficiently derives a dimension hierarchy for a denormalized dimension table. The first information presenting means presents information on a plurality of fact tables included in a multidimensional database to an administrator as first information. The second information presenting means receives a fact table selected by the administrator from among the fact tables related to the first information, and presents information on dimension tables associated with the selected fact table in a star schema to the administrator as second information. The dimension hierarchy deriving means receives a dimension table selected by the administrator from among the dimension tables related to the second information as the one to be processed, calculates a predetermined evaluation value for each column in the dimension table to be processed based on data in each cell belonging to that column, and derives a dimension hierarchy of the columns, which consists of a plurality of hierarchical levels of the respective columns, based on the evaluation values thereof. The third information presenting means presents the dimension hierarchy for the dimension table to be processed derived by the dimension hierarchy deriving means to the administrator as third information. The approval receiving means receives a dimension hierarchy approved by the administrator based on the third information as an approved dimension hierarchy.

Methods and systems are provided for managing knowledge. Users of a knowledge management (KM) service or system may post questions, provide answers or search for answers via a user interface. For a given question, users may collectively construct and refine an answer hierarchy that may include multiple answer paths or solutions to the same question. To this end, users may add, remove and / or modify nodes of the answer hierarchy. The nodes may be of predefined types such as Step, Fork, and Goto. In addition, users may interact with a particular node and provide or modify content specific to individual nodes of the answer hierarchy such as attachments, questions, comments, or references. Given an answer hierarchy, users may select different answer paths or solutions comprising a subset of nodes in the answer hierarchy, for example, based on user-specific fact patterns or requirements.