76 results about "Sentence recognition" patented technology

A system offers potential completions for fragments of text. The system may obtain a text fragment and identify documents that include the text fragment. The system may locate sentences within the documents that include at least a portion of the text fragment, identify sentence endings associated with the located sentences, and present the sentence endings as potential completions for the text fragment.

The present invention relates to an automatic translation and interpretation apparatus and method. The apparatus includes a speech input unit for receiving a speech signal in a first language. A text input unit receives text in the first language. A sentence recognition unit recognizes a sentence in the first language desired to be translated by extracting speech features from the speech signal received from the speech input unit or measuring a similarity of each word of the text received from the text input unit. A translation unit translates the recognized sentence in the first language into a sentence in a second language. A speech output unit outputs uttered sound of the translated sentence in the second language in speech. A text output unit converts the uttered sound of the translated sentence in the second language into text transcribed in the first language and outputs the text.

The invention discloses an external corpus speech identification method based on the deep convolutional neural network. The method comprises steps that S1, speech annotation data and an Internet corpus are obtained; S2, speech signal data is cleaned through utilizing the average energy of the speech signal data; S3, feature extraction and normalization of the speech annotation data are performed;S4, a neural network model is constructed; and S5, to-be-tested speech data is inputted into the constructed neural network model, and identification text data is outputted after identification is completed. The method is advantaged in that the deep convolutional conditional random field model can be constructed based on the speech signal data, compared with a general deep learning model, fewer markup speech data are required, moreover, the cheap large-scale non-markup Internet corpus database is fully utilized to enhance the overall identification rate of sentences, sentence identification accuracy is improved, two processes are integrated, and the end-to-end speech identification method is achieved.

Embodiments of the invention disclose an event relationship graph construction method and apparatus. The method comprises the steps of obtaining event related data; recognizing event sentences in theevent related data according to a trained event sentence recognition model, and acquiring the event sentence recognition model by training the event related data marked with the event sentences as training samples; identifying events and / or relationships in the event sentences according to a trained event relationship identification model, and generating event relationship pairs according to the events and / or relationships, with the event relationship identification model being obtained by training the event sentences marked with the events and / or relationships as training samples; and generating an event relationship graph according to the event relationship pair. The technical problem that the reliability of credit or marketing decision cannot be further improved through the fact logic due to lack of mining and description of the fact logic knowledge in the prior art is solved.

The invention relates to a Chinese text classification method based on BERT and CNN hierarchical connection. The method is mainly used for solving the text classification problems of sentiment analysis, core sentence recognition, relation recognition and the like of Chinese texts. According to the method, hierarchical connection is carried out by using a CNN model and a BERT model to obtain a newmodel BERT-CNN. Due to the fact that the BERT-CNN model is added, sentence features extracted by the BERT model can be further extracted, and more effective sentence semantic representation is obtained. Therefore, in the text classification task, a better classification effect can be obtained.

The invention belongs to the technical field of text content filtration, and particularly relates to a fine-grained semantic detection method of harmful text contents in network. Aiming at an introduced harmful information scene, the method comprises the steps of: constructing a train text set in which independent sentences are used as basic units, thereby establishing a mathematic description of the scene by using a probability topic model; performing information content extraction to a Web page to be detected; performing sentence identification to the text information; calculating a condition probability of each sentence under the model based on the established probability topic model; and accomplishing the fine-grained semantic detection under the set content detection sensitivity. According to the invention, the model construction is hardly affected by the number of the topics, and probability calculation on the sentence and word level is carried out effectively, so that the method is applicable for various application circumstances requiring harmful text content detection; furthermore fine-grained detection to harmful words and sentences of the text content is supported, so that the method improves the detection rate and reduces the misinformation rate effectively, and is beneficial to improving the practicability of text content filtration.

The invention provides a multi-task interaction enhanced electronic text event extraction method, which is a method for extracting an event from an electronic text, combines three tasks of an entity recognition task, a trigger word recognition task and argument role prediction, and additionally introduces an auxiliary event sentence recognition task. Therefore, the attention of the extractor is more focused on the event sentence. According to the method, two sets of parameters are adopted for modeling an entity recognition task and a trigger word recognition task respectively, and informationinteraction is added between sub-tasks, so that a plurality of sub-tasks can be cooperatively trained to realize excellent performance. According to the combined event extraction method, a neural network technology is adopted, a more direct, complete and end-to-end event extraction model is formed, and the effect of an event extraction task is greatly improved.

The invention relates to a case microblog viewpoint sentence recognition construction method of a feature extension convolutional neural network, and belongs to the field of natural language processing. The method comprises the following steps: constructing a case microblog database; annotating comments in the case microblog database to form the training set and the test set of the case microblogcomments; carrying out keyword extraction on a plurality of microblog original texts of the case; taking keywords extracted from the case original text as feature extension and training set case microblog comments, vectorizing the feature extension and training set case microblog comments, and splicing the feature extension and training set case microblog comments to obtain a new vector; splicingthe keywords as feature extension and case microblog comments after vectorization to obtain a new vector as input to train a convolutional neural network, and inputting the test set into the trained convolutional neural network to identify and classify viewpoint sentences. According to the method, the keywords are obtained from the case microblog original text to serve as feature extension, the needed viewpoint sentences are recognized from the obtained public opinion data, and support is provided for subsequent sentiment tendency analysis on the viewpoint sentences.

The invention belongs to the field of language processing and discloses a medical ancient Chinese sentence segmentation method based on Bayesian statistics learning. According to the medical ancient Chinese sentence segmentation method based on Bayesian statistics learning, two tuples and trituples are also added for characteristic attributes or one-tuple, two-tuple and trituple diversified characteristic attributes are combined to obtain multiple groups of experiment data results based on a naive Bayesian method for sentence identification, and finally a best model is obtained; thus, an ancient Chinese sentence segmentation task is achieved. The medical ancient Chinese sentence segmentation method is combined with actual processing text contents, values F of various characteristics in the prior art can be improved by at least 25% by adopting the experiment method, medical ancient Chinese text sentence identification rules are systematically analyzed and concluded, the processing method can be applied to the field of actual traditional Chinese medicine, a medical ancient Chinese text sentence identification corpus is established, and accordingly achievements in scientific research can play a greater role.

The invention discloses a pedestrian monitoring method and device based on intelligent three-dimensional monitoring equipment. The pedestrian monitoring method comprises the following steps: acquiring two pedestrian images with different visual angles in the same scene, calculating by using a stereo matching algorithm to obtain a depth image, respectively taking the depth image and a reference color image as the input of a shallow layer network and a deep layer network, constructing a pedestrian detection network based on convolution feature fusion, and obtaining a pedestrian detection result; carrying out collected audio information, abnormal event keyword sentence recognition by adopting a speech recognition algorithm; and intercepting continuous frames of images according to the keyword sentence recognition result and the audio image synchronous calibration result, performing time domain association representation on fusion features of the continuous frames of images by adopting an LSTM algorithm, and performing passenger flow state and event judgment by adopting a softmax classifier after feature mapping in a full connection mode. The device adopting the algorithm comprises a video image processing and central control unit, a cache module, a storage module, an encryption chip, a network transmission module, a USB interface and a power module.

Provided is an emotional-speech synthesizing device including: a sentence recognition unit that recognizes a sentence that is input; a word emotion determination unit that calculates probability vector of an emotion that is pre-defined for each word that makes up the recognized sentence and estimates the emotion and a rhythm based on the probability vector; and an emotional-speech synthesizing unit. The emotional-speech synthesizing unit calculates in stages degrees of similarity in the emotion and the rhythm between the adjacent words based on context information on the recognized sentence, applies weight to a phoneme candidate corresponding to the each word based on the degrees of the similarity and the probability vector, selects the phoneme candidate that has a minimum target pitch, minimum duration time, a minimum distance value of a target pitch contour, and thus synthesizes an emotional speech that corresponds to the recognized sentence in optimal units.

The invention discloses a Chinese microblog viewpoint sentence recognition feature extraction method based on a self-adaption lifting algorithm. The method comprises the steps that firstly, features related to recognizing a microblog viewpoint sentence are set and recognized, weak classifiers with single features form a strong classifier with a plurality of features, and critical recognition features are selected in the construction process of the strong classifier; finally, an effective subjective sentence recognition feature set and the strong classifier formed by the recognition feature set are output, and effective recognition bases can be provided for recognition of the Chinese microblog viewpoint sentence through the subjective sentence recognition feature set.

The invention discloses a sentence recognizing method and a sentence recognizing device. The sentence recognizing method includes: customizing mapping classifiers, and setting mapping relations between feature vectors and classification results in the mapping classifiers; combining normal classifiers with the mapping classifiers to form weak classifiers; inputting corpus samples into the weak classifiers for classification recognition, wherein the corpus samples contain feature vectors of instantial corpuses; comparing recognized classification results given by the weak classifiers with a standard classification result, if the recognized classification results are consistent with the standard classification result, judging the classification to be correct, and if the recognized classification results are not consistent with the standard classification result, judging the classification to be incorrect; setting weights of corresponding weak classifiers according to an error rate; inputting a to-be-classified sentence into the weak classifiers for classification so as to obtain classification results; making statistics on the weights of the weak classifiers outputting same classification results to obtain probability values of corresponding classification results; taking the classification result with the maximum possibility value as the final recognition result of the to-be-classified sentence. The sentence recognizing method and the sentence recognizing device have the advantage that accuracy in sentence recognition can be improved.

The invention discloses a judgment method and device of document emotion. The method includes: carrying out sentence recognition on a document to obtain a plurality of sentences; extracting all evaluation objects included in the plurality of sentences; obtaining a score, which corresponds to each evaluation object included by each sentence, on the basis of all the evaluation objects included by the sentence and emotion words corresponding to each evaluation object in the sentence; and obtaining emotion types, which are of the document and correspond to each evaluation object, on the basis of the score corresponding to each evaluation object included by each sentence, wherein the emotion types include one of the following items: positive emotion, neutral emotion and negative emotion. The method solves the technical problem that emotion judgment methods in the prior art are low in accuracy of emotion judgment on the document including the plurality of evaluation objects.

The invention discloses a factual implicit emotion recognition method and system based on fusion feature representation, wherein T may be represented by fusing the affective goals at the element level, E may be represented by the factual implicit affective expression at sentence level and B may be represented by the context explicit affective semantic background representation at text level to getthe feature representation of factual implicit affective sentences. In fact-based implicit affective sentences recognition task, the proposed method achieves the highest F1 value of 74.5%, and the highest accuracy rate of implicit affective tendencies discrimination task is 80.5%. Moreover, the proposed method has a good effect in improving the speed of model learning.

A processor-implemented method includes performing speech recognition of a speech signal, generating a plurality of first candidate sentences as a result of the performing of the speech recognition, identifying a respective named entity in each of the plurality of first candidate sentences, determining a standard expression corresponding to the identified respective named entity using phonemes of the corresponding named entity, determining whether to replace the identified named entity in each of the plurality of first candidate sentences with the determined standard expression based on a similarity between the named entity and the standard expression corresponding to the named entity and determining a plurality of second candidate sentences based on the determination result; and outputting a final sentence selected from the plurality of second candidate sentences.

The invention discloses an industry text emotion obtaining method, device and a storage medium, which relate to the technical field of natural language processing big data analysis. The method includes carrying out emotion sentence recognition on clauses in combination with an emotion knowledge base of a target field so as to form an emotion sentence set; according to the constructed affective dependency tree, determining a method of obtaining the scores of affective dependency; the long sentences are segmented into clauses, and the emotional scores of clauses, long sentences and paragraphs are obtained according to the method of obtaining the emotional dependency scores of the target text. Through the practical application test, the industry text emotion calculation method provided by theinvention can better reflect the industry text emotion, and the accuracy rate reaches 85% in the calculation of the long text chapter level emotion. Aiming at the texts of different fields, the invention can quickly construct the domain emotion thesaurus, and realize the fast self-adaptation of the emotion calculation of different industries.

The invention provides a medical text recognition method and a medical text recognition device, and a sentence recognition model training method and a medical text recognition device, which relate tothe medical field. According to the invnetion, by adopting the method of model identification, First, the medical text to be recognized is obtained, after that, the sentences to be recognized in the medical text are structurally extracted, and further obtains a plurality of characteristic words to be recognized of the sentence to be recognized, and then inputting the feature group and the possibleresult (reference result) of the feature words to be recognized into the sentence recognition model at the same time, so that the model outputs the similarity between the feature group to be recognized and each reference result, and finally, the reference result with the highest similarity with the feature group to be recognized is output as the recognition result of the sentence to be recognized, thereby completing the recognition of the medical text.

The invention relates to a Chinese-Vietnamese bilingual multi-document news viewpoint sentence recognition method based on a sentence association graph, and belongs to the technical field of natural languages. Aiming at a Chinese-Vietnamese bilingual multi-document news viewpoint sentence recognition task, the invention provides a viewpoint sentence recognition model combining sentence associationfeatures and semantic features. The method comprises the following steps: constructing a Chinese-Vietnamese bilingual multi-document association undirected graph fusing event elements and emotion elements; obtaining sentence association features of the Chinese-Vietnamese bilingual; obtaining semantic code representation of the sentence; carrying out dimensionality reduction on the obtained semantic codes to obtain sentence semantic features of the Chinese-Vietnamese bilingual; and performing joint calculation by utilizing the sentence association features and the sentence semantic features toobtain viewpoint sentence recognition features, classifying the viewpoint sentence recognition features by adopting a classifier, optimizing the classifier by adopting a binary classification cross entropy loss function, and realizing viewpoint sentence recognition by adopting the optimized classifier. The Chinese-Vietnamese bilingual multi-document news viewpoint sentence recognition method caneffectively improve the accuracy of Chinese-Vietnamese bilingual multi-document news viewpoint sentence recognition.

The invention provides a medical text recognition method and a sentence recognition model training method, which relate to the medical field. The medical text recognition method includes steps of first obtaining a feature code to be recognized corresponding to a sentence to be recognized in a medical text; then the feature code to be recognize and the reference results are input into the train sentence recognition model as input quantities, to determine a similarity between the feature code to be identified and each reference result; at last, the reference result with the high similarity withthe feature code to be recognized is output as the recognition result of the sentence to be recognize, and the tree hierarchy subordination relation of the sentence to be recognized and the pluralityof reference results are simultaneously adopted as input, so that the overall recognition efficiency and accuracy are improved.

The invention discloses a model-based computer software development method, which includes the following steps: S1, the user can input the target requirement statement into the whole development system through the user interaction unit, The system control processing module transmits the target demand statement input in the user interaction unit to the intelligent sentence recognition system, and the original text segmentation processing module in the intelligent sentence recognition system processes the input original text data to obtain the segmented text, which relates to the technical fieldof software development. The model-based computer software development method, can avoid the existing software model building software development methods will waste a lot of people 's time, Realizedby people input the required target-related sentences to automatically extract the reference model, achieved the purpose of shortening the time people build software model, thus greatly facilitatingthe work of software development.

The invention provides a view extraction method and device. The method comprises the steps of obtaining to-be-processed text content; inputting each statement in the text content into a preset subjective sentence recognition model to obtain a subjective sentence in each statement; wherein the subjective sentences are sentences for commenting the comment object; aiming at the subjective sentence, extracting a view quintuple in the subjective sentence; wherein the view quintuple comprises a comment object, a comment attribute, a comment word, a reviewer and comment time; according to the view quintuple, determining the proportion of each type of comments for the comment object and the comment attribute; determining the proportions of various types of comments for the comment objects and thecomment attributes and the subjective sentences as the views corresponding to the text content, so that the subjective sentences in the text content can be extracted firstly, then quintuple extractionand viewpoint analysis operations are carried out on the subjective sentences, the number of quintuples needing to be analyzed is reduced, the accuracy of quintuple analysis is improved, and the requirements for real-time performance and accuracy can be met.

The invention provides a Chinese parallelism sentence recognition method and system. The method comprises the steps that a target dataset is selected, wherein the target dataset comprises multiple types of articles; every two sentences in a current article form a sentence pair, and features of different dimensions are constructed for the obtained sentence pairs; a classification model is constructed according to a random forest method, and the obtained features are used as model input, so that the features are screened according to preset evaluation indexes, and a group of best features are obtained to train the model; and parallelism sentences are extracted from a new test set according to the trained model, and the extracted sentence pairs are combined into sentence chunks for evaluation, so that classification of the sentence pairs is expanded into classification of the sentence chunks. Through the Chinese parallelism sentence recognition method and system, the parallelism sentencesin the articles can be effectively recognized and classified, and therefore an automatic scoring and evaluation system of the articles is actively promoted.

The invention discloses a method and a system for identifying academic viewpoint sentences. The method comprises the steps of firstly, establishing a dominant word database and a hidden viewpoint sentence recognition model; performing dominant word matching on the sentence to be recognized through a dominant word database to obtain a dominant viewpoint sentence matching result; the sentences to berecognized are recognized through the hidden viewpoint sentence recognition model, the hidden viewpoint sentence recognition result is obtained, the technical problem that academic viewpoint sentences can only be recognized from academic texts in a manual mode in the prior art is effectively solved, and the technical effect of efficiently recognizing the academic viewpoint sentences from a largenumber of academic texts is achieved.

The invention discloses a parallel sentence recognition method and system. The method comprises the following steps of: receiving a to-be-analyzed text; carrying out structural processing on the text;obtaining a candidate parallel sentence group from the processed text; extracting parallel sentence features from the parallel sentence group; and carrying out parallel sentence recognition judging on the candidate parallel sentence group by utilizing the parallel sentence features. According to the method and system, text analysis is carried out from abstracted specific features of parallel sentences, so that comprehensive and correct recognition judging results can be obtained.

The present invention relates to a Mandarin monosyllabic word and sentence recognition method, which comprises: inviting an enunciator to pronounce each monosyllabic word, and obtaining K samples of the monosyllabic word with the shortest Baye distance to the known monosyllabic word of the enunciator in a database (i.e., take K samples with the shortest Baye distance to the known monosyllabic word of the enunciator as the best samples of the monosyllabic word, and then abstract characteristics of the K best samples to represent the monosyllabic word, and store the abstracted characteristics in a database). Since K best samples are available to calculate the characteristics for each monosyllabic word, the monosyllabic word recognition capability of the present invention is enhanced greatly. Next, a database of sentences and names is created for the sentences and names to be recognized. In a test of 390 monosyllabic words and 460 sentences and names pronounced by three male and female persons, the successful recognition ratio is 100 percent. In addition, sentences or names can be added to the database at any time, and then the sentences or names can be recognized with the database. Above all, the present invention provides a method for correcting the characteristics of monosyllabic words to ensure successful recognition.

The invention provides a knowledge graph completion method and system based on unstructured information. The method comprises the following steps: identifying entity nodes in missing triple data; obtaining sentences associated with the entity nodes, identifying entity triples in the sentences, and inputting the obtained sentences into a generator to generate free text data; combining the free text data and the structured text data to train a generator, enabling that a discriminator discriminates an entity triple prediction result of the generator according to the entity triple in the sentence, and carrying out adversarial training of the generator and the discriminator; and when the discriminator succeeds in discriminating, adding the entity triples in the sentences into the knowledge graph, scoring through a graph neural network, obtaining the ranking result of the entity triples in combination with the scoring result and the previous node information of the known entity nodes lacking the triples, and completing the knowledge graph. According to the invention, the efficiency and accuracy of knowledge graph completion are improved.

The invention relates to a method for identifying and constructing case microblog opinion sentences using a feature-expanded convolutional neural network, and belongs to the field of natural language processing. The invention includes: constructing a case microblog database; marking comments in the case microblog database to form a training set and a test set of case microblog comments; extracting keywords from a plurality of original microblog texts of a case; The keywords extracted from the training set are used as feature expansion and case microblog comments are vectorized to obtain new vectors; the keywords are used as feature expansion and case microblog comments are vectorized to obtain new vectors as input for training Convolutional neural network, and then input the test set to the trained convolutional neural network to identify and classify opinion sentences. The present invention realizes the acquisition of key words from the original microblog text of the case as a feature extension, identifies the required opinion sentences from the obtained public opinion data, and provides support for the subsequent analysis of the emotional tendency of the opinion sentences.

The invention discloses an international organization science and technology term topic sentence extraction method based on a multi-machine translation engine, and relates to the technical field of information science and knowledge engineering. According to the method, on the basis of a cross-language term knowledge base, by calling various machine translation engines and adopting a natural language chunk processing technology, the technical steps of automatic term sentence recognition, topic statement chunk analysis, translation connection and fusion and the like of the cross-language science and technology text are designed. Therefore, term knowledge point rapid identification, subject sentence automatic detection and high-quality translation fusion generation are carried out on international organization science and technology knowledge, so that subject sentence extraction accuracy and fluency are guaranteed, finally science and technology information issued by an international organization is dynamically monitored, and the knowledge processing requirement of a user for cross-language professional science and technology knowledge is met.