587 results about "Word group" patented technology

To improve efficiency of idea proposal support.SOLUTION: A proposal support system executes: acquiring a specific first word from a co-occurrence network in which each first word of a first word group in a first sentence group including a first field name in at least one information source out of a first information source related to a first field and a second information source related to a second field different from the first field is a node, and a co-occurrence relation between two first words is a link connecting the nodes; extracting company names in the second field relating to the specific first word, from a second sentence group including a second field name and the specific first word from the at least one information source; associating the specific first word, a specific second word, and the company names in the second field, based on appearance information related to the specific second word that co-occurs with the specific first word in a second word group in the second sentence group; and performing output processing of outputting an analysis result in a displayable manner.SELECTED DRAWING: Figure 5

The invention discloses a vision-language pre-training general framework for realizing multi-granularity cross-modal alignment. Image-text pairs are used as training data to be respectively input into an image encoder and a text encoder to obtain coarse-granularity global features and fine-granularity local features; multi-granularity cross-modal alignment is carried out on the obtained global features and local features, and the aligned global features of the image and the text are used for cross-modal pre-training; the multi-granularity cross-modal alignment is composed of a text perception image block selection module and a cross-granularity alignment module; a text perception image block selection module in the step S3 selects the image block with the highest correlation through the correlation between text words or phrases and the image blocks to form a corresponding region or target object, so that more sufficient fine-grained alignment is realized; the cross-modal pre-training target comprises the multi-granularity cross-modal pre-training target constructed in the step S6 and universal image-text comparison loss, image-text matching loss and mask text modeling loss.

The invention provides a cross-modal retrieval model training method and a remote sensing image text retrieval method, and is applied to the technical field of remote sensing cross-modal retrieval. The remote sensing image text retrieval method comprises the following steps: performing feature extraction on a preprocessed remote sensing image by utilizing a trained cross-modal retrieval model to obtain image global features and image local features, and performing feature extraction on a preprocessed target text to obtain text global features and phrase global features; calculating the global similarity between the image global features and the text global features by using the trained cross-modal retrieval model, and calculating the local similarity between the image local features and the word group global features; and carrying out weighted calculation on the global similarity and the local similarity by utilizing the trained cross-modal retrieval model, and screening a matching result between the remote sensing image and the target text based on the weighted similarity to obtain a retrieval result between the remote sensing image and the target text.

The invention provides a question answering method and device based on a large model, electronic equipment and a medium, and relates to the technical field of artificial intelligence, in particular to the technical field of natural language processing, retrieval enhancement generation and intelligent customer service. According to the specific implementation scheme, a question input by a user is received, and then a target knowledge node matched with the question is determined based on word groups included in knowledge nodes in a knowledge base. Based on the target knowledge node, a target document node in the knowledge base is determined, target document content included in the target document node has at least one target text, and the target text is a continuous text to which a word group included in the target knowledge node belongs. And based on the question and the content of the target document, generating an answer to the question by using a large language model. And the data volume input into the large language model is reduced.

A system for deriving a digital representation of an online meeting using contextual data inferred from non-verbal communications. The system processes video streams from meeting participants, including content shared during the meeting. Object detection algorithms identify regions of interest in shared content, such as text collections or graphics. For text collections, layout analysis algorithms determine the structure, enabling extraction of word groups. These word groups are stored with timestamps indicating when they were shared. The system can also detect gestures made by participants, generate textual descriptions of the gestures, and integrate them with a text-based transcript of the meeting. This creates a chronological sequence of both verbal and non-verbal communications. In response to user queries, the system generates responses based on the extracted word groups, gesture descriptions, and their corresponding timestamps, providing comprehensive insights into the online meeting.

The invention provides a multi-modal spoken language pronunciation evaluation method based on a large language model, and belongs to the technical field of spoken language evaluation, and the method comprises the steps: obtaining the audio data of a spoken language tester; performing feature coding processing on the audio data through an acoustic encoder to obtain audio representation information, and mapping the obtained audio representation information into a public space through a modal adaptation layer; setting an audio evaluation instruction, separating the audio evaluation instruction into a plurality of phrases through a word segmentation device, converting the phrases into high-dimensional text features through a word embedding module, and mapping the high-dimensional text features into a public space; splicing the audio representation information in the public space with the high-dimensional text features, carrying out multi-modal feature combination, and constructing an evaluation sequence; and the big language model carries out evaluation prediction on the evaluation sequence to obtain a corresponding pronunciation accuracy evaluation result and a fluency evaluation result. The method does not need to depend on an automatic voice recognition system, and the evaluation efficiency of the audio data is greatly improved while the evaluation process is simplified.

The invention relates to the technical field of digital data processing, in particular to an intelligent triage and guide analysis method, device and system based on a large language model.The method comprises the steps that symptom information of a patient is collected in a text input or voice input mode, and the symptom information and patient identity information are associated; cleaning the collected symptom information and extracting a symptom keyword group; matching the extracted symptom keywords with data in a knowledge graph to determine a to-be-selected disease group; generating a corresponding recommended department table according to the to-be-selected disease group, and arranging the departments according to the priority; and correcting the arrangement sequence of the departments based on the historical treatment records of the patients. Through the steps, the symptom information of the patient can be efficiently collected and processed, and accurate and personalized medical suggestions can be provided. According to the method, the efficiency and the quality of medical services are greatly improved, and meanwhile, the medical experience of the patient is also improved.

The invention relates to a visual-language model cue word evolution generation method based on a genetic algorithm, which comprises the following steps of: randomly generating different cue words under a target task to construct an initial cue word set, preprocessing the randomly generated cue words, and reserving N groups of cue words as an initial population; and designing a multi-dimensional fitness evaluation function to evaluate the performance quality of each group of cue words in a vision-language task, and selecting high-quality cue word individuals based on an elitism strategy. Performing crossover and mutation operation on the selected high-quality cue word individuals by referring to a genetic algorithm to generate a new-generation cue word population; and finally, carrying out iterative optimization for multiple times until a preset termination condition is met, and outputting an optimal cue word set. By the adoption of the visual-language model cue word evolution generation method based on the genetic algorithm, high-quality and diversified cue words can be automatically generated, the performance of a visual-language model is remarkably improved, and meanwhile the visual-language model cue word evolution generation method has good interpretability and adaptability.

The invention relates to the technical field of speech recognition, in particular to a man-machine interaction method and system for speech recognition, and the method comprises the steps: obtaining the frequency spectrum information of a user speech segment, generating a speech spectrogram, constructing a speech library of a user, and carrying out the word segmentation processing, thereby obtaining a sequence of word units; dividing word groups by using a clustering algorithm, and calculating a segmentation coefficient and a stop coefficient of each word group according to a mapping function of an upper envelope line and a lower envelope line of the spectrogram so as to determine an optimal clustering result; real-time input voice segments are compared with reference word groups, and a user confirmation coefficient between the spectrogram of the real-time word groups and the spectrogram of the reference word groups is calculated; and judging whether the real-time input voice segment and the reference voice segment are the same user or not according to the user confirmation coefficient, and completing the human-computer interaction process of voice recognition. According to the method, the user voice library is constructed, and the spectrogram analysis and clustering algorithm is utilized, so that efficient voice recognition and user confirmation are realized, and the user interaction experience is improved.

The invention discloses an intelligent volume adjusting method and system based on environment sound, and the method is applied to a wireless earphone, and comprises the steps: determining a target prompt word according to an initial prompt word registration form submitted by a user; environment sound sampling data collected by wireless earphones are divided into environment noise data and human voice data, and the wireless earphones comprise a left earphone worn on a left ear and a right earphone worn on a right ear; dividing the human voice data into a plurality of groups of character voice data in one-to-one correspondence with the plurality of groups of characters according to voice features in the human voice data; determining target character sound data corresponding to the phrase whose similarity with the target prompt word is greater than a preset threshold value when identifying that the phrase whose similarity with the target prompt word is greater than the preset threshold value exists in the plurality of groups of character sound data; determining the orientation of human voice data corresponding to the target human voice data; and reducing the volume of the left earphone or the right earphone according to the orientation. According to the invention, the use convenience of the wireless earphone is improved.

The invention relates to the technical field of audio auditing, in particular to an intelligent audio auditing method and system based on voice recognition. A sensitive pinyin group and a similar pinyin group corresponding to a sensitive word set are firstly determined, and then voice recognition is performed to obtain a corresponding character field; splitting and recombining the character fields to obtain character phrases, and obtaining character pinyin groups of the character phrases; judging whether a character pinyin group consistent with the sensitive pinyin group or the similar pinyin group exists or not; if the sensitive pinyin group or the similar pinyin group exists, marking the character pinyin group consistent with the sensitive pinyin group or the similar pinyin group as a target pinyin group, and sending the target pinyin group and the corresponding position information to an auditing terminal for re-checking, so that the auditing accuracy is ensured; the auditor only needs to recheck the target pinyin group and the corresponding position information; the auditing efficiency is improved, and the auditing accuracy is higher by adopting a pinyin comparison mode.

The invention relates to the technical field of cross-culture translation, in particular to a cross-culture translation method and system based on context perception fusion, and the method comprises the steps: recognizing a text data set and an image data set, carrying out the paragraph division of the text data set, obtaining a text paragraph set, recognizing an image description text of image data, carrying out the word segmentation of the image description text, and obtaining a text paragraph set; calculating the matching similarity of the text paragraphs and the image description phrases, judging whether the matching similarity is greater than a standard matching similarity or not, if so, recording the matching similarity as a target matching similarity to obtain a target matching similarity set, and obtaining a plurality of associated text paragraph sets according to the target matching similarity set; and translating the associated text paragraph set to obtain an associated translated text, extracting an unassociated text paragraph set from the text paragraph set, and translating the unassociated text paragraphs to obtain an unassociated translated text. According to the invention, the accuracy and efficiency of cross-culture translation can be improved.

Systems and techniques for linguistic learning using automatically cascaded text are described herein. A data model is generated to identify linguistic relationships among respective words and word groups in a source text. The data model is augmented with linguistic properties associated with individual words in the source text. An arrangement of cascaded text is determined based on the linguistic relationships identified in the data model. The arrangement of cascaded text includes horizontal displacement and vertical displacement for the respective words and word groups. Display enhancements to the arrangement of cascaded text are determined wherein the display enhancements are based on user-selected options to show visual aids that identify the linguistic relationships among the respective words and word groups or linguistic properties associated with individual words. The arrangement of cascaded text and the display enhancements are output in a user interface.

The invention discloses a news event classification method. The method comprises the steps of obtaining text data of news events; extracting a plurality of keywords in the text data by utilizing a keyword extraction fusion model to obtain keyword phrases; performing multi-stage similarity calculation on the keyword phrases on the basis of a multi-level keyword library to obtain a similarity calculation result; performing normalization processing on the similarity calculation result, and obtaining a probability distribution result of the news events based on a multilayer sensing network model; and based on the probability distribution result, comparing a difference value of the probability values of the two news events with the highest probability value and the second highest probability value with a preset probability threshold value to obtain a classification result of the news events. According to the method, a more accurate, comprehensive and explainable news event classification result can be provided by comprehensively utilizing different types of data, a fusion model, a multi-level word bank, similarity calculation, normalization processing and flexible classification result processing.

The invention relates to the technical field of data processing, and discloses an intellectual property information analysis and matching system based on big data, comprising: an acquisition unit acquiring user demands and a historical demand set, performing semantic analysis on the user demands to extract demand keywords, and determining importance; sequencing the demand keywords according to the importance to form a current keyword group; the judgment unit is used for comparing the keyword group with a historical demand set and judging whether a historical matching catalog is taken or not; when the historical matching catalogue is judged to be used, determining a current matching catalogue according to the historical matching catalogue and the current keyword group; when it is judged that the historical matching catalogue is not used, determining a current matching catalogue according to the current keyword group; the processing unit performs clustering analysis on the matching catalog to determine a matching display result; and the storage unit stores the current keyword group and the matched display result. According to the invention, the matching efficiency and accuracy are improved, and the problems of low efficiency, low accuracy and the like in traditional intellectual property matching are solved.

The invention relates to the technical field of computers and artificial intelligence, in particular to an industrial economy query word segmentation retrieval method, system and equipment based on context enhancement, large model fine tuning and a vector library and a medium. Firstly, a user inquiry data set is used for performing fine tuning operation on a large model after corpus training is completed, so that the large model can learn and master the context understanding and multi-modal fusion process, and a fine-tuned large model capable of obtaining key phrases from user questions is obtained; the method comprises the steps that firstly, a fine-tuned large model is obtained, then semantic analysis and context completion are conducted on a new question input by a user through the fine-tuned large model, a plurality of new keywords are obtained, finally, a database is inquired according to the new keywords, and an inquiry result is returned to the user. Therefore, an automatic, efficient and accurate database query scheme is provided, the method has the advantages of being low in cost, easy to implement and capable of achieving dynamic self-adaptive optimization, a brand new solution is provided for database query, and practical application and popularization are facilitated.

The invention relates to the technical field of natural language processing, in particular to a dynamic intention chain modeling and causal cleaning method and system under a multi-round dialogue scene. According to the method, through multi-round dialogue data collection and preprocessing, one-hot codes are allocated to each sub-word or word group, intention recognition and sequence modeling are carried out based on the codes, and the intention of the user can be accurately recognized; quick mapping for directly obtaining user intentions from voice input data by skipping text data conversion is established, so that the intention recognition efficiency is improved; according to the method, the causal relationship graph is constructed based on the user intention sequence, the Bayesian network is utilized to perform causal modeling, the causal relationship between intentions is derived, and the probability of reasoning the next intention through the previous intention is calculated, so that the next intention of the user and the interactive operation to be executed are accurately reasoned; and collecting user feedback data to calculate a satisfaction score, and correcting the priority ranking rule of the rule engine based on the score.

The invention relates to the technical field of data identification, in particular to a data security identification method covering physics, semantics and sensitive points, and the method comprises the following steps: processing an obtained first industry review document set and a second industry review document set to generate an industry semantic feature lexicon; performing semantic analysis on each sensitive point information sentence to generate a sensitive point information phrase sequence, taking the sensitive point information phrase sequence as input, performing training through a professional field document to generate a word vector model, and performing similarity calculation on the word vector model to generate a sensitive feature lexicon; based on the industry semantic feature lexicon and a word segmentation tool, keyword segmentation is carried out on the target document, and a semantic security identifier is generated; and performing word segmentation on a target sentence in the target document based on the inactive word bank and the industry semantic feature word bank, and generating a sensitive security identifier in combination with the sensitive feature word bank. By applying the method, effective identification can still be realized under the condition of data adjustment, and semantic representation of the body is enhanced.

The invention discloses a music collaborative playing system and method based on virtual reality. The method comprises the following steps: collecting a scene panoramic video and a scene audio in real time; capturing character activities in a scene panoramic video, identifying a video background, obtaining scene audio frequency, obtaining keywords by using a machine learning model, and constructing a keyword group; obtaining adaptive audio from the database according to the keyword group, judging whether the adaptive audio is replaced or not according to the replay rate of the adaptive audio, and analyzing similar users to obtain the adaptive audio again; the adaptive audio is played, the playing satisfaction degree is obtained according to the character emotion fluctuation condition, and the keyword weight in the audio keyword group is adjusted; real-time character actions are achieved, and when a specific action is recognized, adaptive audio playing is automatically adjusted; the keyword weight of the keyword group of the adaptive audio is adjusted, so that the audio matching accuracy is improved; and the adaptive audio is obtained from the historical playing audio of the similar user according to the keyword group, so that the richness of the adaptive audio is improved.

The embodiment of the invention discloses a physical examination report standardized storage method and device, electronic equipment and a medium. A specific embodiment of the method comprises the following steps: carrying out image enhancement processing on an obtained user physical examination report image set, and then carrying out physical examination item classification processing to obtain a first physical examination item report image set and a second physical examination item report image set; generating an image detection text information set; performing image text recognition on the user physical examination text region image set to obtain a physical examination report text recognition information set; performing text calibration on the physical examination report text identification information set to obtain a physical examination report text calibration information set; generating a report standardized prompt word group set; performing content extraction on the physical examination report text calibration information set to obtain a report extraction information set; and matching and storing the report extraction information set to a user physical examination report storage system. According to the embodiment, the extraction accuracy and effect of the physical examination report information can be improved, the extraction time is shortened, and waste of storage resources is reduced.

The utility model discloses a word spelling teaching aid for Chinese learning, and relates to a Chinese learning tool, in particular to a Chinese polyphonic polysemy word spelling learning tool. The card splicing box comprises a storage box (1), a card splicing box (2) and cards (3), each group of cards comprises a certain polyphonic polysemy, a character card (31) respectively printed with polyphonic polysemy pinyin, a pronunciation card (32) respectively printed with pronunciation different from that of the character card, and a word group card (33) corresponding to the pronunciation card; the storage box is arranged above the card splicing box; a cavity for storing the card (3) is arranged in the storage box (1); and a group of character cards, pronunciation cards and word group cards which are arranged side by side can be accommodated in the card splicing box. The device is a polyphone spelling practice device, and solves the problems that an existing similar device is complex in structure, depends on electric power in use, and is inconvenient to popularize and use.

Gender bias assessment techniques that are available currently only consider two genders viz male and female and is not inclusive of lesbian, gay, bisexual, transgender and queer (LGBTQ) genders. Present disclosure provides method and system for assessing gender fairness of large language models (LLM). The system first receive context aware prompt comprising context and prompt and then converts names present in the context using name anonymization technique. The system then convert many-to-many pronouns present in no-name-context into many to one pronouns which are then used to create counterfactual context for each gender pronoun group. Thereafter, system queries LLM with counterfactual prompt created using counterfactual context to obtain counterfactual response which is then used to create embedding data frame. Further, system trains Gaussian Mixture Model using embedding data frame which is then utilized to calculate cluster distances. Finally, system assess gender bias based on cluster distances and predefined distance threshold.

The invention belongs to the technical field of data mining, and particularly relates to a social media data mining method based on user behavior analysis. The method comprises the following steps: collecting user multi-source behavior data in real time, analyzing unstructured data through natural language processing, extracting key phrases and mapping the key phrases to structured fields; after abnormal data is removed through data cleaning, a behavior sequence is constructed in a time window, a time perception embedded vector is introduced to generate time sequence feature representation, and meanwhile, BERT is utilized to extract behavior content semantic vectors, and features such as behavior types and strength are fused to generate uniform content representation vectors; and finally, fusing the two types of features through a BiGRU-Attention network, and predicting a future behavior probability value of the user to drive personalized recommendation. According to the method, the problems of insufficient behavior semantic motivation capture, difficulty in data isomerism processing, weak dynamic modeling capability and the like in the prior art are solved, and the social media data mining precision is improved.

The invention relates to the technical field of translation processing, and discloses an intelligent machine translation method and device based on a pre-trained large language model.The intelligent machine translation method comprises the steps that a large language model is used for translating a short text corpus to obtain a first translated text, and a phrase set in a translated short sentence is extracted; inputting the combined entity to be tested into the professional knowledge graph for traversal, and judging whether the corresponding translation short sentence has translation ambiguity or not according to a co-occurrence frequency value obtained after traversal; when translation ambiguity exists, ambiguous term words are obtained, replacement phrases are searched out, the ambiguous term words are replaced according to the replacement phrases, and replacement short sentences are obtained; calculating the similarity between each replacement short sentence after back translation and the corresponding original short sentence, and obtaining a corrected translation according to the similarity; repeating the steps until the corrected translations of the original short sentences are obtained, and fusing the corrected translations of the original short sentences into a second translated text; according to the method, translation errors of the terminology can be effectively recognized and corrected.

The invention relates to a voice instruction execution method, system and device, vehicle end control equipment and a storage medium, and relates to the technical field of voice recognition. The method comprises the following steps: identifying a single voice instruction to obtain at least two key phrases, creating a plurality of corresponding sub-tasks based on the key phrases, and adding the sub-tasks into the same to-be-executed task queue; when it is recognized that the to-be-executed task queue comprises the sub-tasks, triggering a preset executor to execute the sub-tasks so as to obtain execution data corresponding to the sub-tasks, and storing the execution data of the sub-tasks to a completed task queue; triggering the associated application to output a task execution result based on the execution data when it is identified that the completed task queue comprises the execution data of the at least two sub-tasks; the number of times of input and recognition of the voice instruction of the user is reduced, the intelligent level of the vehicle end control equipment carrying the voice recognition system is improved, and therefore the use experience of the user on the voice recognition function in the vehicle is improved.

The invention relates to the technical field of legal consultation services, in particular to an intelligent question and answer matching method based on machine learning, which comprises the following steps of: acquiring legal consultation text recognition subject behavior objects and articles, analyzing and carding statement logic in sections, generating a semantic hierarchical table, extracting key phrases and integrating semantic units, and constructing a semantic network to generate a mapping table. Comparing question answer semantic logic to generate a matching corresponding set, reviewing offset revised text to generate a revised set, and rechecking consistency arrangement to generate an intelligent matching result set. According to the method, legal text logic relations are analyzed and processed through semantic layering, question and answer matching precision is optimized, keyword extraction and semantic network construction are combined, semantic consistency recognition and matching flexibility is enhanced, semantic offset and logic inconsistency are corrected, result connection coherence is ensured, mismatching is reduced compared with the prior art, question and answer accuracy is improved, and the method is suitable for popularization and application. Legal consultation is more intelligent and accurate, and a reliable matching result is provided.

To provide an English grammar learning supporting system enabling a user to easily acquire English grammar.SOLUTION: An English grammar learning supporting system 1 includes: a word database 121 which stores English words and coupling graphics whose shapes differ by every word group sharing characteristics of English words in association with each other; a question database 122 which stores order numbers showing word order of the English words in correct English sentence answers and word IDs being identifiers uniquely assigned to the English words, in the word order in questions translating Japanese into English; and a control unit which causes a user terminal 3 to display a guide frame GF, at timing corresponding to an answering state of a user, in a manner of enabling the guide frame GF to become a background of the word card. Therein: the guide frame GF is formed in a shape of a coupling graphic of a case where word cards displaying English words on coupling graphics are connected in order matching word orders of words included in the correct English answers of the Japanese translating into English questions; and the guide frame GF has color / patterns and "shade" which can be distinguished from the word card and the background.SELECTED DRAWING: Figure 14

The invention relates to the technical field of semantic analysis, and discloses an intelligent semantic analysis system and method based on a data large model, and the method comprises the steps: converting an audio file into a text, dividing the text into a character group and a word group, numbering the character group and the word group, and extracting repeated content; training a deep learning model to process the character group and the word group to form a standard text; analyzing the audio to determine a language region, obtaining a dialect database and extracting dialect content; combining the dialects, generating a to-be-analyzed text, and performing semantic analysis and classification; and finally, obtaining the expression meaning of the standard text according to the analysis result. The existing marked dialects are combined and substituted into the standard text, so that potential multiple explanations caused by dialect differences in the text can be covered and revealed, all expression meanings existing in the standard text can be obtained, and the process can ensure that any possible expression meaning is not missed, so that the accuracy of the expression meanings is improved. And the final semantic analysis is more comprehensive and accurate.

The embodiment of the invention provides a model illusion detection method and device, electronic equipment and a storage medium, and belongs to the technical field of text processing and artificial intelligence. The method comprises the steps of obtaining an original text generated by a target model; performing named entity recognition on the original text to obtain a target entity phrase; position query is carried out on the target entity phrase to obtain a position probability, and the position probability represents the probability that any position in the original text is the target entity phrase; based on the position probability, performing illusion scoring on the original text to obtain a text illusion score; on the basis of the text illusion score and the original text, illusion detection is conducted on the target model, a model illusion detection result is obtained, and the model illusion detection result shows that illusion exists or does not exist in the text generated by the target model. According to the embodiment of the invention, the model illusion detection efficiency can be improved.