217 results about "Sentence generation" patented technology

A system and method for generating question-answer pairs is disclosed. The system and method can receive a document. A sentence and / or a further sentence in the document may be identified. A syntactic map for the sentence and / or the further sentence may be generated. Noun phrases and prepositional phrases may be identified based on the syntactic map. Sentence level questions may be generated based on phrases identified using natural language processing (NLP) techniques. Document level questions can also be generated based on syntactic maps generated and NLP techniques.

A method includes preparing a base model using an input model pretrained on at least three languages different from each other and a base vocabulary including words corresponding to two languages among the at least three languages, where the preparing the base model includes constraining the input model to the words included in the base vocabulary; training the base model using a first enhanced training dataset generated from public data, to generate a text summarization model; training the base model using a second enhanced training dataset generated from the first enhanced training dataset, to generate a text generation model; and training the base model using a third enhanced training dataset that is generated using the second enhanced training dataset and the text summarization model, to generate a next sentence generation model.

The invention discloses a method for generating statements from natural languages to SQL (Structured Query Language) based on a large language model agent. The method comprises the following steps of: constructing an LLM (Logical Language Model) guide table mode index for a target database; obtaining a user natural language question and extracting a keyword; searching in an LLM guidance table mode index according to the keyword, and screening related data tables; splicing the table mode information and the keywords of the related data table into a user natural language problem, and constructing an enhanced cue word; based on the large language model intelligent agent, generating SQL statements according to the enhanced cue words and verifying the SQL statements, if the SQL statements succeed in verification, obtaining the required SQL statements, and if the SQL statements fail in verification, performing iterative closed-loop correction on the generated SQL statements through an SQL statement closed-loop correction workflow; if the corrected SQL statement is successfully verified, the required SQL statement is obtained, and if the corrected SQL statement fails and the number of times of correction reaches a threshold value, the suggestions that the natural language problem of the user has defects and is modified are output. According to the method, the accuracy and success rate of SQL statement generation can be improved.

The invention discloses a neural network infrared small target detection method and system based on Mama, and relates to the field of computer vision. The method mainly comprises the following steps: firstly, an image passes through a visual word and sentence generation module to obtain coded visual words and visual sentences; secondly, extracting the features of visual words and visual sentences by using different visual Mama blocks in an encoder; then fusing the features of different levels in a decoder by using a visual Mama block; and finally obtaining a detection result through the detection head. The features are generated and extracted through the visual word and visual sentence module, so that the diversity of the features can be ensured, the feature representation capability of the model is improved, and the Mamba-ISTD-Net can extract the infrared small target visual information with refined granularity; while the linear calculation complexity is maintained, the local features and global features of the infrared small target are fully extracted, and the detection precision and efficiency of the infrared small target can be effectively improved.

The invention relates to an efficient negative sampling and retrieval fused context path inductive relationship prediction method, and belongs to the field of knowledge graph link prediction. The method comprises the following steps: carrying out semantic perception negative sampling; extracting and filtering a path; extracting and filtering a context; performing detailed information retrieval based on a large language model; sentence generation; performing embedding generation; and generating a prediction result and training. According to the method, a more effective negative triple is generated by introducing a semantic perception negative sampling technology, the limitation of traditional random negative sampling is overcome, paths and multi-hop contexts are extracted and filtered from the query triple, long and short entities and relation description are combined into sentences, the multi-hop contexts of a head entity and a tail entity are innovatively sampled, and the sentence extraction efficiency is improved. And irrelevant contents are filtered based on similarity, so that the problem of path scarcity or missing is effectively solved.

The embodiment of the invention provides a data analysis method and system based on retrieval enhancement, and the method can obtain query information in a natural language text form, and retrieves enhancement information in a current domain knowledge base according to the query information, the enhanced information comprises one or more combinations of data definition language annotations, query engine grammar documents, business statement cases and custom domain indexes. And inputting the query information and the enhanced information into a statement generation model, and generating a query statement based on a retrieval result of the statement generation model. And executing the query statement through the database query engine, and obtaining the target data from the database. According to the method, semantic mapping of table names or fields can be established by constructing data definition language statements with annotations and query statement question and answer pairs, and the efficiency and accuracy of natural language query are improved.

The invention discloses a zero sample relation extraction-oriented semantic distillation method. The method comprises the following steps of S1, generating entity and context representation vectors from input sentences; s2, generating entity and context representation vectors of the input relation description; s3, weakening irrelevant features in the context representation through a bidirectional semantic distiller, and strengthening relevant features; s4, performing comparative learning by utilizing enhanced representations of different context representations in the same batch; and S5, performing fine-grained matching on the entity and the context representation of the input sentence and the entity and the context representation in the relation description through a fine-grained semantic matching module. Complex context interference can be effectively dealt with, and generalization ability to different semantic boundaries is improved.

The invention relates to the technical field of retrieval enhancement generation, and discloses a retrieval enhancement generation method based on a reflection mark, which comprises the following steps of: judging whether an answer question needs to be retrieved or not through a large language model, if so, searching a document related to the question, and disassembling the document into a plurality of sentences; generating a reflection mark for each sentence through the large language model, evaluating whether the retrieved document is related to the question and supports question answering, sorting all the sentences according to the reflection marks, and selecting the first Top K sentences from the plurality of sentences as reference contexts of the large language model for question answering by adopting a TOP K algorithm; whether the answer question still needs to be retrieved or not is judged based on the reference context through the large language model, and if yes, the step of obtaining the reference context is repeated until the target large language model considers that retrieval does not need to be conducted again; and the big language model is considered not to need to be retrieved in the reference context target big language model, and the question is answered.

One or more systems, devices, computer program products and / or computer-implemented methods of use provided herein relate to self-reward guided autoregressive sampling for large language models (LLMs). The system can comprise a processor that can execute computer executable components stored in a memory, where the computer executable components can comprise at least one self-reward model. The at least one self-reward model can generate a score for a sentence generated by an LLM, where the score can be based on one or more tokens comprised in the sentence and an attribute associated with the at least one self-reward model. The at least one self-reward model can further alter a text generation process employed by the LLM to generate the sentence, such that respective sampling probabilities of respective tokens comprised in a vocabulary employed by the LLM to generate a new token can be updated by the LLM based on the score.

Automatically generating sentences that a user can say to invoke a set of defined actions performed by a virtual assistant are disclosed. A sentence is received and keywords are extracted from the sentence. Based on the keywords, additional sentences are generated. A classifier model is applied to the generated sentences to determine a sentence that satisfies a threshold. In the situation a sentence satisfies the threshold, an intent associated with the classifier model can be invoked. In the situation the sentences fail to satisfy the classifier model, the virtual assistant can attempt to interpret the received sentence according to the most likely intent by invoking a sentence generation model fine-tuned for a particular domain, generate additional sentences with a high probability of having the same intent and fulfill the specific action defined by the intent.

Method includes populating fake value for each of entities, to generate string of fake entity values that correspond to entities; inserting sentinel token between adjacent fake values included in the string to generate first input data; generating, by natural language generation model, natural language sentences based on first input data, natural language sentences including one or more fake values from the string; analyzing natural language sentences to determine whether any fake value from the string is missing; based on the fake value missing, summarizing, using text summarization model, natural language sentences to generate text summary; concatenating the text summary with the fake value, to generate second input data; and generating, by a next sentence generation model, additional natural language sentence using the second input data, the additional natural language sentence including the fake value. Additional natural language sentence is combined with natural language sentences to generate final natural language sentences.

To efficiently conduct interviews.SOLUTION: An information processing device comprises a text generation unit and an output unit. The text generation unit generates response texts to user utterances and question texts for asking the user preset interview items using a text generation model for generating responses to the input texts on the basis of expert knowledge related to an interview subject and interview knowledge related to interviewing skills. The output unit outputs the response texts and question texts generated by the text generation unit to the user.SELECTED DRAWING: Figure 2

To allow for outputting appropriate information according to the emotions of a user.SOLUTION: An action control system disclosed herein comprises a collection unit for collecting condition information representing the condition of a user viewing an art piece, and an output control unit configured to provide control so as to cause an avatar controlled by a control device with a sentence generation model to take an action according to the condition information collected by the collection unit.SELECTED DRAWING: Figure 2

The present invention generates, with a minimum effort by a worker, a question sentence to be inputted to a generative AI device and acquires an NC program which realizes a desired operation of an industrial machine.　This program creation assistance device assists in creating a numerical control program which causes an industrial machine to operate. The program creation assistance device comprises: a machine information acquisition unit which acquires machine information regarding the industrial machine; a question sentence generation unit which acquires operation information regarding desired operation contents of the industrial machine, and generates a question sentence for a generative AI device by combining the operation information with the machine information acquired by the machine information acquisition unit; a transmission unit which transmits, to the generative AI device, the question sentence generated by the question sentence generation unit; a reception unit which receives a first response result from the generative AI device; and a presentation unit which presents the first response result.

To compare information transmission in an unspecified field with information extracted from another information source.SOLUTION: A non-transitory computer-readable recording medium storing a program for operating a computer, the program causing a processor of the computer to execute: receiving designation of a text; and extracting, from an information source available to an artificial intelligence system, at least one source text including a description similar to a designated text that is the designated text. Generating a prompt including an instruction to output information about the source sentence and an instruction to compare the extracted source sentence with a specified sentence, and presenting, to a user, the information about the source sentence and a result of the comparison between the specified sentence and the source sentence, based on an answer obtained from a large scale language model provided by an artificial intelligence system by inputting the prompt to the large scale language model.SELECTED DRAWING: Figure 10

The invention belongs to the crossing field of artificial intelligence and barrier-free assistance technology, and particularly discloses a sign language translation method and device based on a large language model.The method is applied to a terminal and comprises the steps that key points are extracted from sign language video frames; based on the coordinates, types and global indexes of the key points in the sign language video frames, obtaining fusion features through feature fusion; based on the fusion features corresponding to the sign language video frames in the frame window, time sequence modeling and classification are carried out, and a word prediction result of the frame window is obtained; a sliding window voting and deduplication mode is adopted, the candidate word sequence is preprocessed, a preprocessed word sequence is obtained, and the candidate word sequence is formed by connecting word prediction results of all frame windows in series according to a time sequence; based on the preprocessed word sequence, word error correction and statement generation are carried out by calling a large language model, and a sign language translation result is generated. According to the invention, sign language video translation can be realized for the terminal, and the accuracy and fluency of the translation result can be guaranteed.

The invention discloses a structured query statement generation method and device, equipment and a storage medium, belongs to the field of data processing, and can accurately generate SQL statements according to natural statements. The method comprises the steps of obtaining a to-be-generated statement, and determining intention information in the to-be-generated statement; based on a preset sub-graph determination rule, determining a first sub-graph structure corresponding to the intention information in a preset database graph structure, the database graph structure being formed based on a plurality of first graph nodes and an association relationship between the first graph nodes, and the first graph nodes being determined based on structure information of a plurality of associated databases; a plurality of preset second graph structures are obtained, the second graph structures with the structural similarity with the first sub-graph structures larger than a first threshold value are determined as third graph structures, and different second graph structures are generated based on different types of preset SQL statements; and determining statement generation parameters in the first sub-graph structure based on the third graph structure, so as to generate a target SQL statement based on the statement generation parameters.

To provide a business assistance system etc. which can efficiently acquire information relating to an insurance product.SOLUTION: A business assistance system comprises: a reception unit which receives inquiry from a user; a user identification unit which identifies the user on the basis of transmission source identification information used for transmission of the inquiry; an inquiry content acquisition unit which acquires an inquiry content of the user, on the basis of the inquiry; an inquiry content recognition unit which recognizes the acquired inquiry content; an inquiry destination selection unit which selects an inquiry destination for inquiring the recognized inquiry content, from a communication destination list stored in advance; a sentence generation unit which generates a sentence for informing the selected inquiry destination of the recognized inquiry content; and a sending unit which clearly indicates that it is the inquiry from the identified user, and sends the generated sentence to the selected inquiry destination.SELECTED DRAWING: Figure 3

The invention discloses a prompting method and system for enhancing the implicit reasoning ability of a large language model, and the method comprises the steps: receiving an original context, and generating a forward reasoning problem and a backward reasoning problem based on the sentence of the original context; the forward reasoning problem and the backward reasoning problem are screened, problems which are related to original context semantics and have logic values are reserved, and an intermediate reasoning problem set used for auxiliary reasoning is generated; combining the intermediate reasoning problem set with the original task input, and constructing an enhancement prompt; and inputting the enhancement prompt into the large language model and obtaining an answer. According to the method, a potential causal logic chain is complemented for the model, so that explicit and implicit information can be fused more sufficiently, the reasoning accuracy and robustness are improved, and the implicit information identification capability is enhanced; and the reasoning ability can be enhanced only by prompting the project without retraining or fine tuning of the model.

To allow for outputting appropriate information according to the emotions of a user.SOLUTION: An action control system disclosed herein is configured to detect occurrence of an event related to circumstances of a user, and output information to the user according to the detected event using an avatar controlled using a sentence generation model.SELECTED DRAWING: Figure 2

The application provides a power industry science and technology front and application article personalized recommendation method and system, a storage medium and an electronic device, and relates to the field of personalized recommendation.In the application, the related field coefficient is used to screen sentences related to the personalized knowledge graph in the article to be recommended, and a new text after information filtering is generated; the title of the article to be recommended and the new text are used as the input of the BERT model, the title representation and the sentence representation are obtained respectively, the attention mechanism at the sentence level is introduced, the attention weight of the title representation to each sentence representation is obtained, the weighted sentence representation is obtained, and finally the classifier is input.The application uses the constructed personalized knowledge graph to screen sentences, preliminarily filters out irrelevant information, introduces the sentence-level attention model, filters the preliminarily screened sentences again, greatly reduces the interference of information irrelevant to the theme, and improves the performance of the classification model, so that the personalized recommendation result with high accuracy is quickly generated.

To allow for outputting information appropriate for an event that has occurred.SOLUTION: An action control system disclosed herein comprises a detection unit for detecting occurrence of a predetermined event, and an output control unit configured to provide control so as to cause an avatar controlled by a control device with a sentence generation model to output information to a user according to an event detected by the detection unit.SELECTED DRAWING: Figure 2

In implementations of systems for generating templates using structure-based matching, a computing device implements a template system to receive input data describing a set of digital design elements. The template system represents the input data as a sentence in a design structure language that describes structural relationships between design elements included in the set of digital design elements. An input template embedding is generated based on the sentence in the design structure language. The template system generates a digital template that includes the set of digital design elements for display in a user interface based on the input template embedding.

In a fault cause identification support device, an error message acquisition means acquires an error message from an IT system. A configuration information acquisition means acquires configuration information of the IT system. A device state acquisition means acquires state information of each of devices forming the IT system based on the error message and the configuration information. A question sentence generation means generates a first question sentence including the error message, the configuration information, the state information, and an instruction sentence for instructing analysis of an error cause. A response means inputs the first question sentence to a large-scale language model and acquire one or more candidates of the error cause as an answer. Thus, the fault cause identification support device capable of supporting in identifying a cause of a fault in the IT system is provided.

To provide an information processing device capable of generating a more appropriate sentence based on a user input sentence, and a program.SOLUTION: In a document generation system in which a user device and a response generation device are respectively connected to a sentence generation device via a network so as to enable mutual communication, the document generation device being an information processing device comprises: an input acquisition unit which acquires a plurality of user input sentences respectively associated with a plurality of items from a user; and a sentence generation unit which generates a first instruction sentence instructing to generate a first sentence of a format including at least a plurality of sentences tailored to the plurality of items on the basis of the plurality of user input sentences, and inputs the first instruction sentence to a trained model to generate the first sentence.SELECTED DRAWING: Figure 5

A factuality verification apparatus includes an input module configured to receive a written document, a filtering module configured to filter an untargeted sentence, which does not correspond to a target for factuality verification, a selection module configured to respectively divide a plurality of individual facts of one sentence into different sentences to generate atomic sentences respectively corresponding to the plurality of individual facts and select a core atomic sentence corresponding to a sentence including the fact-related content associated with the subject among the atomic sentences, and a verification module configured to perform factuality verification on whether the core atomic sentence is true or not, through comparison of the core atomic sentence and an evidence document capable of checking whether content of the core atomic sentence is true or false.

The invention discloses a statement generation method, which comprises the following steps of: receiving load characteristic information and database mode information, and obtaining characteristic parameters according to the load characteristic information and the database mode information; determining the statement type of the generated statement according to the feature parameters; and calculating a state transition probability according to the load feature information. In the embodiment of the invention, the load type required by the user can be determined according to the database mode information and the load feature information, and the SQL statement of the load of the determined type is generated according to the load feature information. The styles of the SQL statements of the load generated by the method are more, so that the styles of the SQL statements of the load are enriched, and the load generation mobility is improved.

The application relates to an artificial intelligence-based sentence generation model updating method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: obtaining a query sentence triggered by an object identifier, generating a plurality of candidate sentences associated with the query sentence through a sentence generation model; selecting a candidate sentence selected through the object identifier as a positive sample, and selecting a candidate sentence not selected as a negative sample; respectively extracting features of the positive sample and the negative sample to obtain semantic representations corresponding to the positive sample and the negative sample; constructing a contrast loss function between the positive sample and the negative sample based on the semantic representations of the positive sample and the negative sample; and updating and training the sentence generation model through the contrast loss function until a stop condition is reached, thereby obtaining an updated sentence generation model. The method can improve the accuracy of the generalization result of the sentence generation model in generating a result associated with the query sentence.

This application proposes a social text enhancement method based on multimodal retrieval and keyword extraction, comprising: S1, extracting keywords from sample sentences of different categories using a category keyword extraction algorithm; S2, training a sentence generation model using an RNN model combined with a self-attention mechanism, and controlling the sentence generation direction of the generation model according to the keywords of the corresponding categories of the training samples; S3, inputting the original sentence into the generation model to generate a first generated sentence with text enhancement; S4, based on a multimodal retrieval algorithm, determining whether the first generated sentence contains keywords from the keyword file; if so, finding the keyword to be replaced in the first generated sentence, and retrieving synonyms of the keyword to be replaced for replacement, thereby generating multiple second generated sentences with data enhancement. This application has the effect of controlling the generation direction and number of generated sentences of the generation model.