207 results about "Word target" patented technology

An apparatus that calculates a confidence measure of a target word string specified in a recognition result includes: an alternative candidate generator which generates an alternative candidate word string in the position of the target word string; a classifier training unit which trains a classifier which is configured to discriminate between the target word string and the alternative candidate word string; a feature extractor which extracts a feature value representing an adjacent context in the position of the target word string; and a confidence measure calculator which determining whether the true word string in the position of the target word string is the target word string or the alternative candidate word string by using the classifier and the feature value and calculates a confidence measure of the target word string on the basis of the determination result.

The invention discloses a commodity target word oriented emotional tendency analysis method, which belongs to the field of the analysis processing of online shopping commodity reviews. The method comprises the following four steps that: 1: corpus preprocessing: carrying out word segmentation on a dataset, and converting a category label into a vector form according to a category number; 2: word vector training: training review data subjected to the word segmentation through a CBOW (Continuous Bag-of-Words Model) to obtain a word vector; 3: adopting a neural network structure, and using an LSTM(Long Short Term Memory) network model structure to enable the network to pay attention to whole-sentence contents; and 4: review sentence emotion classification: taking the output of the neural network as the input of a Softmax function to obtain a final result. By use of the method, semantic description in a semantic space is more accurate, the data is trained through the neural network so as to optimize the weight and the offset parameter in the neural network, parameters trained after continuous iteration make a loss value minimum, at the time, the trained parameters are used for traininga test set, and therefore, higher accuracy can be obtained.

The invention relates to an emotion classification method and system, a storage medium and equipment. The method comprises: encoding a context by using a position word vector and multi-head self-attention; encoding the target word by using a bidirectional gating circulation unit and multi-head self-attention; fully extracting semantic information and position information of long and short sentences; meanwhile, interactively splicing the context semantic information and the target word semantic information for low-order fusion; performing position coding on the basis of low-order fusion by using a capsule network; and performing high-order fusion on the information after low-order fusion by using a multi-head interaction attention mechanism, averagely pooling the high-order fusion information, and splicing the averagely pooled high-order fusion information with the averagely pooled target word semantic information and the averagely pooled context semantic information to obtain target feature representation. Compared with the prior art, the context semantic information, the target word semantic information and the position information are fully fused, and the emotion classification accuracy and efficiency are improved.

The invention relates to the field of artificial intelligence, discloses a method, device and equipment for obtaining a supervision recognition result in multiple modes and a storage medium, and solves the problem of semantic similarity matching of current business supervision terms and business products. The method comprises the following steps: creating a knowledge graph; processing the knowledge graph according to a first preset rule, a second preset rule and the entity relationship file to obtain an entity and an entity relationship; updating the knowledge graph according to the entities and the entity relationship to obtain a target knowledge graph; analyzing the target knowledge graph and the training text through an encoder to obtain fused to-be-processed information; performing random masking processing on the fused to-be-processed information according to a preset strategy to obtain training data; performing word embedding vector processing and self-made force mechanism processing on the training data to obtain a target sentence vector and a target word vector; and calculating a weighted average value of the semantic cosine similarity and the character string similarity ofthe target sentence vector and the target word vector according to a preset weight ratio to obtain a supervision and identification result.

The invention relates to a specific target emotion classification task based on a graph neural network. The method comprises the following steps: acquiring a data set and initializing a BERT model; obtaining a one-dimensional feature vector of each target word through a BERT model; inputting the feature vector of the target word into a graph convolutional neural network model; constructing a network topological graph, calculating an adjacency matrix, obtaining three features of nodes in the network topological graph in three modes according to the adjacency matrix, introducing relation classification tasks,wherein the whole model is divided into two stages and two tasks in classification, and the two tasks are emotion polarity classification of target subjects and relation classificationbetween the target subjects respectively. According to the method, the graph neural network is adopted to compose a plurality of subjects appearing in sentences and process a plurality of targets at the same time, so that the cognitive law of judging emotion polarity by human beings is better met, the effect of a model is ensured, meanwhile, a relationship classification task is introduced for auxiliary classification, and the classification accuracy is further improved.

The invention relates to the field of artificial intelligence, and provides a text recognition method, device and equipment and a storage medium, and the method comprises the steps: carrying out the classification of a collected word data set according to industry types, and building a plurality of types of word banks; classifying the plurality of types of word banks according to business types toobtain a plurality of candidate business type word banks, and sorting the plurality of candidate business type word banks according to priorities to obtain a plurality of initial target business wordbanks; performing identification processing on a target image based on the plurality of initial service type word banks through a preset image text identification model to obtain a text prediction result corresponding to the target image and a target service type word bank; obtaining a target word in the target business type word bank, and establishing a data structure tree according to the target word; and obtaining a word with the highest matching degree with the target word from the data structure tree, and outputting the word with the highest matching degree as a text recognition result.By adopting the scheme, the accuracy of text recognition can be improved.

Disclosed are a speech recognition device and a speech recognition method, which perform speech recognition by executing an artificial intelligence (AI) algorithm and / or a machine learning algorithm provided therein, and which can communicate with other electronic devices and an external server in a 5G communication environment. According to an embodiment, the speech recognition method includes setting an additional wake-up word target capable of activating a speech recognition function in addition to a preset wake-up word, generating a plurality of additional wake-up word utterances formed on the basis of the additional wake-up word target being uttered under various conditions, learning a wake-up word recognition algorithm by using each of the spoken utterances of the additional wake-up word to generate an additional wake-up word recognition algorithm, and executing the additional wake-up word recognition algorithm upon receiving a select word uttered by a user to determine whether to activate the speech recognition function.

The invention provides an intention understanding method and device, and the method comprises the steps: recognizing a target word slot label corresponding to a composition word in a target text, andgenerating a target generalized text of the target text according to the target word slot label; matching the target generalization text with a plurality of preset generalization templates, and determining candidate generalization templates according to the matching degree; calculating the semantic similarity between the target generalization text and the candidate generalization template, and determining the target generalization template according to the semantic similarity; and obtaining a template intention of the target generalization template, taking the template intention of the targetgeneralization template, and generating an intention understanding result of the target text according to the template intention, the target word slot label and the composition words corresponding tothe target word slot label. Therefore, intention understanding of related information is realized based on a generalization processing mode, dependence on a large number of annotation training samplesis avoided, annotation workload and difficulty of a sample person are reduced, and intention understanding modes are enriched.

A vocabulary dictionary storing unit for storing a plurality of words in advance, a vocabulary dictionary managing unit for extracting recognition target words, a matching unit for calculating a degree of matching with the recognition target words based on an accepted voice, a result output unit for outputting, as a recognition result, a word having a best score from a result of calculating the degree of matching, and an extraction criterion information managing unit for changing extraction criterion information according to a result of monitoring by a monitor control unit are provided. The vocabulary dictionary storing unit further includes a scale information storing unit for storing scale information serving as a scale at the time of extracting the recognition target words, and an extraction criterion information storing unit for storing extraction criterion information indicating a criterion of the recognition target words at the time of extracting the recognition target words. With the change in the extraction criterion information, the vocabulary dictionary managing unit increases or decreases the number of the recognition target words. This makes it possible to improve a recognition performance without the need for a user to carry out a troublesome operation.

The embodiment of the invention provides an intention classification method and device, electronic equipment and a computer readable storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: obtaining a request text; performing entity feature extraction on the request text to obtain a first text containing a target query parameter; inputting the first text into a pre-trained comparison model, and performing matrix multiplication on the first text and a reference word embedding matrix in the comparison model to obtain a plurality of target word embedding vectors; performing classification processing on the target word embedding vector by utilizing a pre-trained intention classification model to obtain a target word embedding vector containing an intention category label and an intention classification probability value; performing matching processing on the first text by utilizing a pre-trained intention matching model to obtain an intention matching value; and obtaining intention classification data according to the intention matching value and the intention classification probability value. According to the embodiment of the invention, accurate classification of user intentions can be realized, and the intention classification accuracy is improved.

The invention provides a natural language query statement processing method and device and electronic equipment. The method comprises the steps of obtaining a natural language query statement of a user; according to a preset data table and a preset keyword table, performing target word extraction and identifier labeling on the natural language query statement to obtain a target word combination, namely a plurality of target words and attribute identifiers and position identifiers corresponding to the target words; according to the attribute identifiers and the position identifiers corresponding to the multiple target words, performing text reconstruction on the multiple target words by utilizing a target text reconstruction rule corresponding to the target word combination to obtain a target query statement; and converting the target query statement into a database execution language so as to carry out information query. According to the method and device, the natural language query statement of the user can be converted into the target query statement through the target word extraction and the rule-based text reconstruction process, then the target query statement is converted into the database execution language, and the accuracy of information query can be greatly improved through two times of conversion.

The invention relates to the technical field of natural language processing, and provides a target emovement analysis method and device, a target emovement analysis model training method and device, acomputer storage medium and electronic equipment. The target sentiment analysis method comprises the steps of for each word in a to-be-tested statement, obtaining a word vector of the current word, obtaining a text vector used for representing global semantic information of the current word, and obtaining a position vector used for representing position information of the current word in the to-be-tested statement; encoding a word vector, a text vector and a position vector of each word in the to-be-tested statement to obtain a semantic vector of the to-be-tested statement; obtaining a targetvector corresponding to a target word in the to-be-tested statement; and predicting the sentiment polarity category of the target word according to the semantic vector and the target vector. According to the technical scheme, the prediction accuracy of the sentiment polarity category of the target word can be improved.

The embodiment of the invention provides a text similarity calculation method and device, electronic equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the steps of: acquiring a to-be-calculated original text; performing word segmentation processing on the original text by utilizing a pre-trained text word segmentation model to obtain a plurality of text word segments; performing position identification on each text word segment by utilizing a pre-trained target word library model to obtain a target position of each text word segment; performing coding processing on each text word segment according to the target position to obtain a text word segment vector; inputting the text word segment vector into a pre-trained comparison model so as to perform matrix multiplication on the text word segment vector and a reference word embedding matrix in the comparison model, thus obtaining a target word embedding vector; and performing similarity calculation on the plurality of target word embedding vectors to obtain a similarity value between every two text word segments. According to the embodiment of the invention, the correlation between the text word segments can be calculated more accurately.

The invention provides a word vector generation method and device and electronic equipment, and relates to the technical field of natural language processing. The method comprises the steps: obtaininga to-be-processed statement, and obtaining a target word in the to-be-processed statement; obtaining context information of the target word; determining a context-independent word vector corresponding to the target word according to a preset mapping relation between segmented words and word vectors; determining a context semantic vector corresponding to the context information according to a preset mapping relation between characters and word vectors; and generating a target word vector corresponding to the target word according to the context-independent word vector and the context semanticvector. Therefore, through the word vector generation method, the reliability and universality of natural language understanding are improved, and the calculation complexity of word vector generationis reduced.

The invention provides a text generation method and device, and relates to the technical field of natural languages. The invention provides a text generation method and device. The method comprises the steps of determining a first segmented word in a first text corpus; under the condition that the first segmented word is matched with the preset feature, replacing the first segmented word with a feature mark corresponding to a preset feature to obtain a second text corpus, obtaining a first word vector corresponding to a first segmented word in the first text corpus, the first segmented word and the first feature vector corresponding to the feature mark in the second text corpus; inputting the first word vector and the first feature vector into the text generation model at the moment, obtaining the output target word vectors of the target semantic structure, and then obtaining the target text corpus according to the target word vectors. The target text corpus obtained by the embodimentof the invention comprises the required and complete target semantic structure, and the first text corpus does not limit the obtaining mode, so that the problems of high similarity and poor extensioneffect between the extended target text corpuses are avoided.

The invention provides an emotion classification method. The emotion classification method comprises the steps of obtaining a word embedding matrix corresponding to a context and a word embedding matrix corresponding to a target word; according to the word embedding matrix corresponding to the context, the word embedding matrix corresponding to the target word and the first semantic activation model, obtaining context representation with enhanced target word meaning and target word representation with enhanced context semantics; obtaining context representation after semantic selection according to the context representation of the target word semantic enhancement, the target word representation of the context semantic enhancement and the semantic selection model; according to the semantic integration model, extracting syntactic representation in a syntactic dependency tree corresponding to the target sentence; and obtaining an emotion classification result corresponding to the target word according to the context representation, the syntax representation and the second semantic activation model after semantic selection. Compared with the prior art, the semantic information related to the target word in the context is fully captured, and the relationship among the context, the target word and the syntax is comprehensively considered, so that the accuracy of sentiment classification is improved.

The invention relates to the field of computers, in particular to the technical field of artificial intelligence, and discloses a text classification method and device, electronic equipment and a storage medium; the method comprises the steps: obtaining to-be-recognized text information, inputting the text information into a trained first text classification model, and obtaining a target word vector matrix; performing semantic mining processing on each target word vector to obtain a corresponding semantic feature, and finally obtaining a target prediction classification result based on each semantic feature, wherein the first text classification model is obtained by performing parameter adjustment based on a first loss value and a second loss value, the first loss value is an error value between the prediction classification result and the actual classification result, and the second loss value is an error value between the two prediction classification results. According to the invention, two loss values are used for adjusting parameters of the first text classification model, so that a predicted classification result of the first text classification model approaches to an actual classification result and another predicted classification result, and the classification accuracy of the model is further improved.

The invention relates to the technical field of deep learning, and provides a semantic recognition model training method and device based on comparative learning and a medium, and the method comprises the steps: carrying out the synonym replacement of each first target vocabulary in an original sentence text, and obtaining a first processed sentence text; carrying out antonym replacement on each second target vocabulary in the original sentence text to obtain a second processed sentence text; and training an initial semantic recognition model through the original sentence text, the first processed sentence text and the second processed sentence text to obtain a target semantic recognition model. The objective of the embodiment of the invention is to provide a comparative learning method capable of identifying small semantic changes, and the problem that positive and negative samples are well distinguished in the prior art is improved, so that the model can learn richer knowledge, and the model training effect is improved.

The invention relates to the field of intelligent decision making, and discloses a text classification method, which comprises the steps of cleaning an original text to obtain a target text, performing word segmentation on the target text to obtain a word segmentation set, extracting a keyword set, and obtaining a position information set and a part-of-speech information set of the keyword set; converting the keyword set, the position information set and the part-of-speech information set into a keyword vector set, a position information vector set and a part-of-speech information vector set,carrying out vector splicing to obtain a target word vector set, identifying semantic information of the target word vector set by utilizing a semantic identification model to obtain a word semantic information set, and identifying the text category of the target text according to the text semantic information of the target text and the word semantic information set. The invention also relates toa blockchain technology, and the target text can be stored in a blockchain node. The invention further provides a text classification device, electronic equipment and a readable storage medium. According to the invention, the text classification accuracy can be improved.

The invention relates to an information element joint extraction method and system based on deep learning. The method comprises the following steps of: converting an input target text into a target word vector representation by utilizing a pre-training language model and a bidirectional long-short-term memory network; enumerating all text spans of each sentence in the target word vector, and obtaining a target text span vector representation based on the target word vector representation; constructing a text span graph network corresponding to the co-reference relationship, the entity relationship and the event structure relationship, and transmitting and updating text span vector representation; and performing multi-task classification on each updated text span vector representation. According to the method, the target text is converted into the target word vector representation, and the text span vector representation integrating a local context and a global context can be learned, so that span graph networks corresponding to different relationships can be constructed and the text span vector representation is updated, and then task classification of the text span vector representation is realized; and the association degree among the tasks is improved, so that the performance of the tasks is improved.

The invention discloses a synonym pushing method. The synonym pushing method comprises the steps of obtaining interview questions; configuring a first preset number of keywords of answers corresponding to the interview questions; pre-training based on the super-large word vector model to obtain a target word vector model; constructing a word vector matrix according to the target word vector modelto obtain a word-index file; constructing a binary tree based on all word vectors; traversing the binary tree, querying a first candidate word vector from the binary tree, wherein the distance betweenthe first candidate word vector and the keyword is greater than a preset distance threshold, and constructing a priority queue; carrying out de-duplication on the first candidate word vector in a priority queue; obtaining a second preset number of target word vectors sorted in the top in the deduplicated priority queue; and pushing a second preset number of synonyms for the user to select based on the target word vector and the word-index file. The invention further provides a synonym pushing device, electronic equipment and a storage medium. According to the invention, synonyms can be quickly pushed to users.

The embodiment of the invention discloses a method and device for generating information. A specific embodiment of the method comprises the steps of obtaining a comment information set; for the comment information in the comment information set, extracting target words from the comment information, the extracted target words forming word groups, wherein the target words comprise evaluation objectwords and evaluation words; for the phrases in the at least one formed phrase, inputting the evaluation object words and the evaluation words in the phrase into a pre-trained prediction model to obtain a prediction result, determining whether the evaluation object words and the evaluation words in the phrase are matched or not based on the prediction result, and if yes, classifying the phrase intoa phrase set; and generating viewpoint information based on the phrases in the phrase set. According to the embodiment, generation of viewpoint information with relatively high readability is realized.

The invention relates to a text enhancement method and system based on word interpretation, and the method comprises the steps of obtaining a to-be-tested text, and obtaining an interpretation sentence of a target word in the to-be-tested text; then preprocessing a to-be-tested text, for a text classification task taking the target word as a label, setting an explaining sentence of the target word as a label, and for a text classification task not taking the target word as the label, adding the explaining sentence of the target word into the text; and finally, training the natural language classification model through the preprocessed text, so that the problems that the model training effect is poor and the accuracy rate is low when the text is classified and words with insufficient related corpora such as new words exist in the text are solved, and the accuracy rate of the model is improved.

The embodiment of the invention discloses a state detection method and device, computer equipment and a storage medium, and belongs to the technical field of computers. The method comprises the stepsof determining a relationship matrix corresponding to a state description text, wherein the relationship matrix comprises a first association weight between every two words in the state description text; based on the relation matrix, performing feature enhancement processing on an original word vector matrix corresponding to the state description text to obtain a target word vector matrix; performing detection processing according to the target word vector matrix to obtain a detection result; and performing feature enhancement on the words by adopting the association weights between the words,so that the obtained target word vector matrix not only can represent the features of the words, but also can represent the association degree between the words and the association degree between thewords and the target type state. Therefore, the state detection is performed according to the target word vector matrix, and the association relationship between the characteristics of the words andthe words is fully considered, so that the considered factors are richer, and the accuracy of the state detection can be improved.

The invention discloses a Chinese word vector generation method based on similar contexts and reinforcement learning. The problems that an existing Chinese word vector generation method carries out prediction on the basis of the relation between adjacent contexts of target words and the target words, the situation that although some words in Chinese are adjacent, semantics are irrelevant is not considered, and the expression quality of word vectors is not high are solved. The method comprises the following steps of: selecting a corpus, and preprocessing the corpus so as to construct a Chinesecorpus; carrying out similar context discovery on the Chinese target words to obtain similar contexts related to semantics of the Chinese target words; and constructing a Chinese word vector reinforcement learning framework, and performing reinforcement learning to obtain word vector representation of the Chinese target word. According to the method, the problem that adjacent Chinese words are irrelevant can be solved, and high-quality Chinese word vectors are generated.

Embodiments of the invention provide a search method and apparatus, an electronic device and a computer readable storage medium. The method comprises the steps of obtaining a target search term; according to a pre-established word segmentation library, performing word segmentation processing on the target search word to obtain a first word segmentation result; searching data matched with the firstword segmentation result in a pre-established database; wherein words in the word segmentation library are target words extracted from historical search records of a user in a machine learning mode,and the target words are used for describing search targets of historical search words. Therefore, according to the scheme, the problems that based on an existing word segmentation method, when searching is carried out, the searching time is long, and the searching result accuracy is low can be solved to a certain extent.

The embodiment of the invention provides a word vector generation method and related equipment. According to the method, the importance degree of the word vectors in the words can be added in the process of converting the word vectors into the word vectors through an attention mechanism, the weighted average of the word vectors is combined with the original word vectors, the final word vectors areobtained, and the expression capacity of the word vectors can be effectively improved. The method comprises the steps of obtaining a target word, wherein the target word is a word of a word vector tobe generated; determining an initial word vector of the target word, a word vector of each word in the target word and a global variable through a training model; determining the weight of each character in the target word through the global variable and the character vector of each character; and determining a target word vector of the target word according to the word vector of each word, the weight of each word in the target word and the initial word vector of the target word.

The invention is suitable for the technical field of artificial intelligence, and provides a data relationship extraction method and device, electronic equipment and a storage medium, and the method comprises the steps of obtaining a text sequence from a to-be-processed text, and obtaining a related dependent word of each word in the text sequence and a semantic relationship between each word andthe related dependent word based on a syntax dependency tree; generating a target word embedding vector corresponding to the to-be-processed text according to each word, the related dependent word ofeach word and the semantic relationship; and inputting the target word embedding vector into the deep convolutional neural network, performing entity relationship information extraction on the to-be-processed text based on the target word embedding vector, and obtaining a target entity relationship which is output by the deep convolutional network and has a set prediction probability value. The invention can be applied to a smart medical scene, the extraction convenience and accuracy of the entity relationship in the related information are improved, and the construction of a smart city is promoted.