321 results about "Noun" patented technology

A machine translation method includes receiving a source text string and identifying any named entities. The identified named entities may be processed to exclude common nouns and function words. Features are extracted from the source text string relating to the identified named entities. Based on the extracted features, a protocol is selected for translating the source text string. A first translation protocol includes forming a reduced source string from the source text string in which the named entity is replaced by a placeholder, translating the reduced source string by machine translation to generate a translated reduced target string, while processing the named entity separately to be incorporated into the translated reduced target string. A second translation protocol includes translating the source text string by machine translation, without replacing the named entity with the placeholder. The target text string produced by the selected protocol is output.

Methods, systems and computer programs for automatic, highly accurate machine comprehension of a plurality of segments of free form unstructured text in a natural language. The system answers a plurality of complex, free-form questions asked in a natural language, based on the totality of input text. The system further uses a multi-dimensional data model to measure the total effects of actions / verbs acting on various unique nouns present in the input text. The system may convert the questions into another multi-dimensional data model and may then compare the two data models in program memory to derive the answers to the posed questions. The system may then automatically detect unknown words and optionally look them up in digital information sources, such as online dictionaries and encyclopedias, to fill in the gaps in knowledge to answer the questions with expert-like reliability.

A method of correcting errors in a speech recognition system includes a process of searching a speech recognition error-answer pair DB based on a sound model for a first candidate answer group for a speech recognition error, a process of searching a word relationship information DB for a second candidate answer group for the speech recognition error, a process of searching a user error correction information DB for a third candidate answer group for the speech recognition error, a process of searching a domain articulation pattern DB and a proper noun DB for a fourth candidate answer group for the speech recognition error, and a process of aligning candidate answers within each of the retrieved candidate answer groups and displaying the aligned candidate answers.

A system and method for estimating a metric of a proposed product from a document describing the product in an arbitrary degree of detail provides accuracy comparable to a function point assessment of a fully detailed document by parsing a document into a plurality of lines in an outline form and assigning a hierarchy to each line. Values or weights are determined for each of a relatively small plurality of key words in accordance with the key word and the hierarchy level of the line in the parsed document in which the key word is found. An adjustment is made based on a category of the key word such as nouns and verbs and the hierarchy level of the lines as compared with the number of hierarchy levels found in the document. Further adjustments may be made for subject matter domain of the product and the source or author of the document.

The invention relates to a method for relationship classification with an LSTM and CNN joint model based on location. The method includes the steps of (1) preprocessing data; (2) training word vectors; (3) extracting location vectors; acquiring the location vector feature and high-dimensional location feature vector of each word in a training set, cascading the word vector of each word with the high-dimensional location feature vector thereof to obtain a joint feature; (4) building a model for a specific task; encoding contextual information and semantic information of entities by use of bidirectional LSTM; outputting the vector of the location corresponding to the marked entities, inputting the output to CNN, outputting two entity nouns and their contextual information and relational wordinformation, and inputting the entity nouns and their contextual information and relational word information into a classifier for classification; (5) training the model by use of a loss function. The method does not need to manually extract any features, the joint model does not need to use additional natural language processing tools to preprocess the data, the algorithm is simple and clear, and the best effect at present is achieved.

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.

User profiles are created based on taxonomic nouns related to documents accessed by the user. The profiles can be leveraged to create lists, such as mailing lists and lead lists, to target content, such as offers, to persons most likely to be interested in the content. A database of the profiles is queried based on nouns describing content to be promoted. The profiles that satisfy the query are used to generate a list. The invention can be used to create any type of list, such as mailing lists, lead lists, lists of related content, lists of related users, lists of categorized content, and the like.

Unstructured electronic documents are classified for profiling and targeting users for additional relevant content. Behavioral data is gathered from user activity, and user documents and actions are categorized. Profile information is combined with collaborative and editorial data to provide users with credible information regarding products. Author-generated document classification information is analyzed and assigned a first taxonomic noun to characterize the document. User-generated tags characterizing a portion of the document are assigned a second taxonomic noun. Search terms that resulted in the user accessing the document are identified and assigned a third taxonomic noun. Attributes related to how the document was accessed are evaluated and assigned a fourth taxonomic noun. The document is processed using pattern rules to extract a fifth taxonomic noun. The taxonomic nouns are aggregated to determine term vectors representing the document, and the document is categorized using the term vectors, the taxonomic nouns, or the author-generated classification.

A verbal classification system and method for the efficient sending and receiving of desired information is described. Web-site proprietors (314) and Internet marketers provide a host server (320) with descriptive data about their web-sites and messages (336) according to a predefined verbal hierarchy of nouns, verbs and modifiers. An Internet user creates a role or roles (340) that describes desired information according to the same predefined verbal hierarchy of nouns, verbs, and modifiers and transmits that role (340) to the host server (320), provides the Internet user with desired web-site addresses (336) and message by matching the role (338) to web-site and message descriptive data (336). The Internet user may access a web-site using one or more roles (340) and the web-site may provide the user with custom content by matching content (338) to the Internet user's role and roles (340). The host server (320) collects role information (340) from Internet users and web-sites to generate marketing and demographics data (346) about Internet users and web-sites (328). Internet user privacy is protected because role data (340) is encrypted and contains no personal identification information (348).

The invention discloses a classified processing method of internet public information. The method comprises the following steps: selecting a classified public information text as a training text, and parsing words; selecting and screening nouns and verbs, acquiring feature words by extraction, vectorizing the training text, then acquiring a PCA transformation feature matrix, a BP neural network model, and a decision tree rule; performing dimension reduction on vectors of the vector matrix of the public information text to be classified by the PCA transformation feature matrix, and transforming the vectors by the BP neural network model to obtain an output vector which has the same number of dimensions as the classified number, and then performing matching by the decision tree rule, and determining that the public information text to be classified belongs to the public information category marked by the rule if the matching is successful. As the PCA transformation converts a feature word space related to a high dimension into a low-dimensional orthogonal feature space, the disadvantage of inaccurate classification is solved; meanwhile, the decision tree rule is used for classification without data similarity comparison so that a plurality of data sources can be processed in a short time.

The invention relates to a data searching method and system based on semantic analysis, and belongs to the technical field of information searching. The method comprises the steps that keywords input by a user are divided into words with different parts of speech through a word dividing library, for results after the words are divided, geographical positions to be searched can be determined according to a geographical noun library if the parts of the speech are geographical names, otherwise matching is carried out by entering a phrase word library, query is further carried out based on keywords corresponding to the matched record, and words input by the user are regarded as the keywords of the inquire if matching is failed. The data searching method and system based on the semantic analysis carries out the semantic analysis based on a large number of behaviors of the user, can carry out good analysis on non-concrete nouns, verb-object phrases, modification phases and other vocabularies with generalization properties, can automatically select new words needing to be put into a library at the same time, is few in searching repeated process, high in searching efficiency and relatively precise in researching result and can meet demands of clients.

This invention utilizes a concept called color, which implies a variation, and applies it to natural language attributes like verbs and nouns. The verb color is defined as a role or operation in which the field participates. The noun color is defined as a form of the field. The auxiliary verb color is defined as a path to the field from a known object reference. The noun color may be defined by the user or may be determined, based on the object state in which the field resides. Equations are made generic, by making the colors of the fields parametric. The equivalence of methods and equations was established, and a method might be invoked, as if it was a rule. Similarly, processes involving several methods, and other rules, may be described by rules. A special kind of classes called conceptual classes were invented, which can project a subset of the fields of a class, as well as group and reorder a particular field found in several classes. Several objects called collaboration objects, may interact with each other in several cycles, and in each cycle, several methods of the objects are invoked. Each method may view the collaboration objects in a predefined order called collaboration sequence. Temporary variables created during processing may be stored in a global or local table, and may be assigned user defined or state based nouns. By utilizing all the mechanisms defined above, rules may be specified and evaluated in a generic manner.

A method of storing service description documents in a computerised storage system in which each document is associated with at least one verb ontological node (204) and at least one noun ontological node (254), each verb ontological node having one or more links to other verb ontological nodes and each noun ontological node having one or more links to other noun ontological nodes whereby the verb nodes form a verb space (200) and the noun nodes form a noun space (250) and a method of retrieving service description documents from a plurality of service description documents stored in this way comprising the steps of: controlling a user interface to request from a user at least one verb request term (405) and at least one noun request term (410), associating the or each verb request term (405) with a corresponding verb node (204) and the or each noun request term (410) with a corresponding noun node (254), comparing the or each corresponding verb node (204) with the or each verb node (212, 214) associated with each of the stored service description documents, comparing the or each corresponding noun node (254) with the or each noun node (262, 266) associated with each of the stored service description documents, and selecting for retrieval zero or more of the stored service description documents on the basis of the comparison steps and controlling the user interface to inform the user of the selected documents to enable the user to retrieve one or more of the selected documents.

A device or computer implemented program for accurately and automatically obtaining general-purpose information regarding the usage difference between a plurality of synonyms and quasi-synonyms, such as the types of words with which the synonyms and quasi-synonyms are often used, is provided with: means for receiving the input of a plurality of words; means for extracting sentence data including an inputted word from a corpus; means for analyzing the sentence structure of the sentence data and extracting nouns that are in a grammatical relationship with the inputted word included in the sentence data; means for extracting the nodes representing the nouns and the nodes representing the semantic category of the noun from a thesaurus and forming a directional graph for each inputted word; means for comparing a plurality of directional graphs and extracting the difference nodes; and means for outputting the extracted difference nodes as information relating to the usage difference of the inputted words.

The invention relates to the field of artificial intelligence, and provides a text processing method and device and a model training method and device, and the method comprises the steps: obtaining target knowledge data which comprises a first noun entity, a second noun entity and a contact between the first noun entity and the second noun entity; processing the target knowledge data to obtain a target knowledge vector; processing a to-be-processed text to obtain a target text vector, the to-be-processed text comprising the first noun entity; fusing the target text vector and the target knowledge vector according to a target fusion model to obtain a fused target text vector and a fused target knowledge vector; and processing the fused target text vector and / or the fused target knowledge vector according to a target processing model to obtain a processing result corresponding to the target task. According to the technical scheme, the accuracy of the target task processing result by the target processing model can be improved.

A text semantic analysis method and system can realize semantic analysis of text data base on lexical level and sentence level. Aiming at the semantic analysis at the lexical level, the invention firstly adopts an improved word segmentation algorithm to solve the problem that English words are segmented only by spaces. Secondly, based on word segmentation, TF-IDF modeling is performed to obtain weight value; Then the text is vectorized by weighting and summing the weight value and the word vector trained by Word2Vec, and finally the document similarity is solved. At the same time, the invention considers the contribution degree of the vocabulary to the document content and the semantic status to calculate the similarity degree of the document, the result has higher accuracy, and provide agood foundation for subsequent text clustering. The present invention extracts subject-predicate object structure based on text segmentation, part-of-speech tagging, syntactic analysis and dependencyrelation for sentence level semantic analysis. The invention realizes the extraction of subject-predicate-object structures of various sentence types in all aspects, and realizes the noun expansion function, which is more consistent with the manual extraction result.

The invention discloses a human-computer interaction intelligent question answering method based on a cloud platform, which comprises the following steps: acquiring voice information input by a user;performing voice-to-text conversion on the voice information to obtain text information; performing word segmentation processing on the text information to obtain keyword extraction results; using a machine learning algorithm to classify the obtained keyword extraction results so as to obtain classification results; using a natural language processing algorithm to perform keyword expansion on verbs and nouns in the keyword extraction results and taking out a result with the largest similarity from the results of the keyword expansion in each of the verbs and nous, wherein all the results forma keyword expansion sequence; and performing fuzzy matching in a local database according to the classification results and the keyword expandable sequence. The human-computer interaction intelligentquestion answering method based on the cloud platform in the invention can solve the technical problem of low interaction accuracy caused by inaccurate word segmentation, inaccurate keyword expansion,and inaccurate extraction of answers in the existing human-computer interaction question answering system.

The invention discloses a multi-dimensional geographic scene identification method fusing geographic region knowledge. The method comprises the steps of preprocessing images in a database to obtain satisfied geographic scene images; obtaining object region image blocks by utilizing a method for quickly searching for object regions in the images; pre-training the obtained object region image blocks of the geographic images by using a deep convolutional neural network, performing an accurate adjustment process until the performance of the deep convolutional neural network of the scene images is no longer improved, and fusing feature matrixes into output eigenvectors; pre-establishing a geographic entity noun keyword dictionary by acquired entity noun data in geographic scene classification, performing word segmentation on target identification result data to obtain key words in a target identification result, and establishing text features; and fusing the text features and multi-dimensional image features into eigenvectors as inputs, realizing cross-media-data identification classification, and realizing scene classification fusing geographic entity information.

The features that are presented to an evolutionary algorithm are preprocessed to generate combination features that may be more efficient in distinguishing among classifications than the individual features that comprise the combination feature. An initial set of features is defined that includes a large number of potential features, including the generated features that are combinations of other features. These features include, for example, all of the words used in a collection of content material that has been previously classified, as well as combination features based on these features, such as all the noun and verb phrases used. This pool of original features and combination features are provided to an evolutionary algorithm for a subsequent evaluation, generation, and determination of the best subset of features to use for classification. In this evaluation and generation process, each combination feature is processed as an independent feature, independent of the features that were used, or not used, to form the combination feature. In this manner, for example, a particular phrase that is generated as a combination of original feature words may be determined to be a better distinguishing feature than any of the original feature words and a more efficient distinguishing feature than an unrelated selection of the individual feature words, as might be provided by a conventional evolutionary algorithm. The resultant best performing subset is subsequently used to characterize new content material for automated classification. If the automated classification includes a learning system, the evolutionary algorithm and the generated combination features are also used to train the learning system.

The invention relates to a system suitable for a blind person to surf the Internet, comprising an automatic server news downloading system and a client application system. A server system can realize real-time news downloading, store the news into a server and realizes real-time updating; and a client system can correspondingly respond to voice input of the blind person by virtue of voice recognition and synthesis and outputs a voice. For special case of the blind person, the system realizes three core functions of the Internet, namely information acquisition, knowledge learning and interaction. By applying the system provided by the invention, the blind person can effectively listen to news by virtue of the Internet, listen to an electronic books, inquire encyclopedia nouns and report and listen to a post, wherein the news listening system can support the blind person to sequentially listen to news in programs, and the blind person can pay close attention to interested news by keyword search and related news search.

The invention relates to a method and a system for managing multiple on-line users. The method comprises the following steps: obtaining the user password; judging whether the use password comprises awakening nouns; if the use password comprises the awakening nouns, extracting the voiceprint characteristic of the use password, and comparing the obtained voiceprint characteristic with the voiceprint characteristic in a voiceprint password base; determining whether the user can execute the next operation according to the comparison result. The system comprises an obtaining unit, a voiceprint password base, a judging unit, an extracting unit, a comparing unit and a determining unit. The invention can realize the managements of instant login and authority identification for multiple special users.

The invention discloses an RNN-based automatic picture description generation method. A deep web which is well trained in advance is firstly used for image feature extraction; non-noun and non-verb components are removed for words in the sentence; an LSTM network is finally used for joint training on the image features and lexical features; during the sentence generation process, a sentence formed by nouns and verbs is generated through the inputted image and the well-trained LSTM network; and then, through large corpus on the network, the final outputted sentence is generated. Automatic recognition can be realized, a digital image uploaded by the user is understood, and a natural sentence understood by a human being is generated.

The present invention provides a technique, by speech input, to specify a location, even when a name or the like of the location is not accurately known, while driving. An in-vehicle apparatus mounted in a vehicle has a speech recognition function, and stores information shown on a signboard and a noun substitute in storage. If speech inputted from a speech input device includes a noun substitute and further includes at least a part of the information shown on the signboard, a response is decided according to the information shown on the signboard, and the decided response is outputted from an output device.

A method and a server for extracting a topic and evaluating suitability of the extracted topic are disclosed. The topic extraction server includes a text preprocessing unit configured to extract noun from a document group and remove stopword from the extracted noun, a keyword extraction unit configured to calculate a weight of a noun and extracting a keyword representing the document group, a seed selection unit configured to calculate a weight of the extracted keyword and select a seed, an initial clustering unit configured to generate one cluster including the selected seed and a keyword shown by several times in a sentence including the selected seed, and a cluster combination unit configured to extract a topic group.

A system and method for matching similar master data using associated behavioral data, for example transactional data. Matching accuracy is significantly increased. Master data may be thought of as nouns. Behavioral data is associated with master data as verbs are related to nouns. Specifically, behavioral data is data associated with an action taken by a master data object. Behavioral data may include temporal and non-temporal data. Temporal data for example may include time and / or duration that a behavior occurred. Non-temporal data may include a physical location, a product, a phone number or any other quantity other than time. Using behavioral data in duplication testing adds another dimension to de-duplication that is not achieved through comparison of master data object fields alone. For example, a duplication test on two similar “person” records with similar “name” field values is improved by comparing associated “behavior” data, (phone numbers called by each “similar person”).

The invention discloses an identification method for a webpage information related region. The identification method for the webpage information related region includes: 1) establishing a regional information body; 2) extracting metadata information and text of crawled webpage information and dividing the extracted information title and text into words; 3) analyzing geographical name pronouns which express places in the words, judging whether there are reference relationships between the geographical name pronouns and the previous geographical terms, if so, replacing the geographical name pronouns with the corresponding geographical terms; 4) analyzing non-standard geographical name words in the words, and replacing the non-standard words with standard words; 5) analyzing the regional information of the relative position based on the regional information body to acquire an accurate geographical name word; 6) judging the analyzed webpage information based on the regional information body, and classifying the webpage information into a successfully matched region. The identification method for the webpage information related region greatly improves the identification accuracy for the webpage information related region.