27 results about "Query expansion" patented technology

The invention discloses a retrieval enhancement generation method and system based on multi-dimensional reordering, and relates to the technical field of information retrieval, and the method comprises the following steps: S1, constructing a tensor index, a keyword index and a compression abstract; s2, driving a Qwen large model to execute query expansion and hypothetical answer generation through a dual-task query processing template; s3, mixed retrieval is executed through a semantic and keyword double-path index, and two-stage reordering is executed by using an improved DistilBERT model and a Qwen large model; s4, performing multi-dimensional evaluation through sub-problem decomposition; s5, iterating according to the sequence of the sub-questions, and gradually generating answers to the sub-questions; and S6, executing multi-dimensional evaluation correction, and generating a final answer. According to the method, the limitation of a traditional retrieval enhancement generation technology on information matching precision, context correlation evaluation and answer generation quality is overcome, and an efficient and accurate solution is provided.

The invention relates to the technical field of information retrieval, and discloses a term standardization query method and system based on a large language model, and the method comprises the steps: receiving original query, dialogue history and multi-modal input; analyzing the multi-modal input to generate an extended context; candidate aliases are extracted through natural language processing, and a domain term knowledge base is inquired in combination with a term list; adjusting the priority weight according to the user role and the real-time context; constructing a structured prompt including dialogue history, original query, extended context, candidate schemes and task instructions, and inputting a large language model to generate a replacement decision; updating the knowledge base based on user interaction data and model feedback; historical query records are stored to assist in follow-up reasoning. According to the invention, the picture, the PDF and the voice input are analyzed through the multi-modal context enhancement module, the extended context containing the term list and the semantic clue is generated, comprehensive context support is provided for term standardization, and query processing is ensured to adapt to various enterprise scenes.

The invention relates to the technical field of digital archive management and information retrieval, and discloses a digital archive multi-modal data semantic enhancement fusion retrieval method and system.The method comprises the steps that a policy cycle time axis and a policy term evolution graph are constructed, tense logical reasoning is conducted on archive seals, and permission effectiveness evolution is derived; the temporal permission feature vector and the content semantic vector are fused to generate a multi-modal representation vector, and cross-policy-cycle semantic enhancement retrieval is realized by combining query expansion and temporal permission filtering, so that the problems of missing detection and misjudgment of policy and regulation archives in seal permission historical evolution and term cross-cycle retrieval are solved.

A system and method for personalizing health information content provided to a user. The techniques include obtaining user profile data associated with the user in response to a query submitted by the user. The user profile data includes health data of the user and application use data by the user. The techniques further include expanding the query into a plurality of queries based on the health data of the user, performing semantic searches for the plurality of queries, ranking results from the performed semantic searches, aggregating the results from the performed semantic searches based on the ranking to generate an aggregated health information response, and providing the aggregated health information response to the user.

The invention discloses a query expansion method and system for power grid knowledge search, and the method comprises the steps: obtaining an initial query word of a user, and retrieving an initial document set from a power grid document set; segmenting the document into text fragments by using a preset power grid entity boundary dictionary, and calculating a context comprehensive weight of each fragment; calculating the semantic similarity between the initial query word and each fragment through the word vector, and screening out candidate expansion words in high-similarity fragments; a weighted co-occurrence graph is constructed based on the co-occurrence relation of the candidate expansion words, and edge weights are determined by accumulating context comprehensive weights of co-occurrence fragments; and combining the word vector similarity with the initial query word to carry out comprehensive sorting, selecting a word ranked in the top as an expansion word, and combining the expansion word with the initial query word to form an expansion query.

The invention discloses a building material semantic retrieval method based on a vector database, and relates to the technical field of building informatization management, and the technical scheme is as follows: the method comprises the following steps: obtaining building material data and carrying out cleaning and standardization processing, extracting material fields and business attribute information, and establishing synonym mapping information or material and product name mapping information; carrying out blocking and vectorization processing on the processed building material data, and storing the processed building material data to a vector database; performing query understanding on the natural language query of the user, complementing missing key fields and performing unified name expansion; performing query expansion based on the processed query field to generate a retrieval expression; and performing vector retrieval and keyword retrieval, fusing and reordering to obtain a target retrieval result, and performing extended re-retrieval when the result is empty or insufficient. The method has the beneficial effects that the retrieval accuracy, the recall capability and the result availability under the scenes of different names, unified names and material substitutive names of the building materials can be improved, and the empty result rate and the manual screening cost are reduced.

A system and method for personalizing health information content provided to a user. The techniques include obtaining user profile data associated with the user in response to a query submitted by the user. The user profile data includes health data of the user and application use data by the user. The techniques further include expanding the query into a plurality of queries based on the health data of the user, performing semantic searches for the plurality of queries, ranking results from the performed semantic searches, aggregating the results from the performed semantic searches based on the ranking to generate an aggregated health information response, and providing the aggregated health information response to the user.

This invention provides a method for injecting industrial knowledge based on retrieval enhancement, belonging to the field of industrial knowledge technology. This invention establishes a time-series data flow matrix by collecting multi-source heterogeneous data, constructs a time-series causal knowledge graph using Granger causality tests, establishes an industrial knowledge document library with a hybrid index structure, performs time-series-aware query expansion on the query input, performs a hybrid retrieval of dense vectors and sparse inverted indexes and cross-encodes and reorders the data, associates the refined documents with the time-series causal knowledge graph to extract event evolution paths and fuses them to generate a time-series enhanced knowledge representation, inputs it into a time-series knowledge enhancement model to generate answer text containing fault analysis and prediction suggestions, injects it into an industrial decision support system after quality assessment, and stores feedback data for continuous model optimization. This solves the technical problem of industrial knowledge retrieval being disconnected from real-time temporal status, resulting in a lack of time-series causal reasoning ability in the generated answers.

The invention provides a query processing method which can be applied to the technical field of artificial intelligence. The query processing method comprises the steps that an original query text is acquired and preprocessed; inputting the preprocessed original query text into a pre-trained target model to obtain an entity of the original query text; performing standard normalization on the entity of the original query text; performing query expansion and reconstruction on the entity after standard normalization to obtain executable query representation; based on a target query template determined by the intention of the executable query representation, filling placeholders of the target query template with the executable query representation to obtain a target query statement for retrieval; wherein the target model is a double-architecture model containing a generation path and a compatible path. The invention further provides a query processing device and equipment, a storage medium and a program product.

This invention discloses a semantic retrieval method for building materials based on a vector database, belonging to the field of building information management technology. The technical solution includes: acquiring building material data and performing cleaning and standardization processing; extracting material fields and business attribute information; establishing synonym mapping information or material-name mapping information; segmenting and vectorizing the processed building material data and storing it in a vector database; performing query understanding on user natural language queries, completing missing key fields and expanding general terms; expanding the query based on the processed query fields to generate retrieval expressions; performing vector retrieval and keyword retrieval, fusing and re-ranking to obtain the target retrieval results, and performing expanded re-retrieval when the results are empty or insufficient. The beneficial effects of this invention are: improving the retrieval accuracy, recall, and result usability in scenarios involving synonyms, general terms, and material abbreviations of building materials, while reducing the empty result rate and manual screening costs.

The embodiment of the application discloses a research report generation method, which comprises the following steps: obtaining original text data from at least two data sources, generating overlapping text slices by using a sliding window slicing algorithm, and storing the text slices in a vector database through a text embedding model; performing multi-dimensional weighted scoring on each data source, combining a time decay factor and a Bayesian historical accuracy factor to calculate a final confidence score; providing a three-layer tree hierarchical report template, supporting user selection or semantic recommendation; performing query expansion and multi-stage semantic retrieval for an analysis problem, detecting semantic contradiction by using natural language reasoning, and performing conflict resolution based on confidence and correlation, and finally generating a complete research report by assembling the answer content with traceability annotation generated by a large language model after weighted fusion. The technical scheme of the application can realize quantitative evaluation of data confidence, has a multi-source data conflict resolution mechanism, and the content generation is traceable.

The invention discloses an intelligent operation and maintenance disposal behavior monitoring method based on artificial intelligence, and the method comprises the steps: firstly, carrying out the judicial evaluation of the value of an initial result through preliminary retrieval and content validity judgment, and avoiding unnecessary query rewriting; only if necessary, the screened high-quality key text fragments are utilized to carry out conditional and constrained query expansion on the original alarm problem. Through the controlled expansion mode, the semantic deviation risk caused by unconstrained rewriting of a general model is avoided from the source, and it is ensured that secondary retrieval can more accurately focus on the core intention of original warning. Therefore, key information with higher signal-to-noise ratio and stronger correlation can be provided for subsequent scheme generation, and the accuracy and practicability of the finally generated disposal scheme are remarkably improved.

The invention discloses a water surface ship target re-identification method based on deep learning, and the method comprises the following steps: S1, obtaining a training data set of a water surface ship image, and carrying out the fogging processing of the image in the training data set according to a proportion; s2, using a YOLO model to carry out training on the training data set after fog adding processing; s3, training on the fogged training data set by using a visual auxiliary re-identification network with a defogging module; s4, performing detection by using the trained YOLO model to obtain a prediction frame, and performing prediction frame de-weighting by using a non-maximum suppression method based on pseudo-overlapping rate joint confidence; and S5, cutting each input frame of image according to a prediction frame, inputting the cut image into the visual angle auxiliary re-identification network to extract ship features and visual angle features, and identifying the ship identity by using a visual angle auxiliary adaptive query expansion method to obtain a re-identification result. The method significantly improves the precision of ship target re-identification, and still has good robustness in a foggy environment.

The present application belongs to the field of data encryption, and relates to a privacy protection multi-dimensional range query method and device and a storage medium, wherein the method comprises the following steps: a data owner generates a key required for an entire ciphertext range query; the data owner constructs an HHCB tree by using a layered Hilbert encoded ciphertext; a data user encodes a query range according to a layered Hilbert encoding mode, and generates a query threshold by using a hash function to perform hash mapping on a range encoding set; and a cloud database performs query expansion on the HHCB tree by using the query threshold, and returns a query result obtained to the data user. Compared with the prior art, the present application introduces a layered Hilbert encoding mode, reduces a cumbersome encryption and decryption calculation process during query, improves query efficiency, introduces an approximate encoding division method in index construction, improves traversal efficiency, and is more suitable for an actual cloud ciphertext database scene.

The application discloses a retrieval enhancement generation method and system based on multi-dimensional reordering, relates to the technical field of information retrieval, and comprises the following steps: S1, constructing a tensor index, a keyword index and a compressed abstract; S2, driving a Qwen large model to perform query expansion and hypothetical answer generation through a double-task query processing template; S3, performing mixed retrieval through a semantic and keyword double-path index, and performing two-stage reordering by using an improved DistilBERT model and the Qwen large model; S4, performing multi-dimensional evaluation through sub-problem decomposition; S5, iteratively generating sub-problem answers according to the sub-problem order; and S6, performing multi-dimensional evaluation correction to generate a final answer. The application overcomes the limitations of traditional retrieval enhancement generation technology in information matching accuracy, context relevance evaluation and answer generation quality, and provides an efficient and accurate solution.

The present application relates to a Chinese-Vietnamese cross-language query expansion method based on retrieval enhancement and knowledge distillation, and belongs to the technical field of natural language processing. The present application injects the thinking chain generation ability of a large-scale language model and the retrieved external knowledge into a multilingual pre-training model with fewer parameters through knowledge distillation and retrieval enhancement, thereby improving the thinking chain generation ability of the multilingual pre-training model. Compared with query expansion, cross-language query expansion can improve the reasoning and generation ability of a multilingual pre-training model in a low-resource language scenario. The present application plays an important role in Chinese-Vietnamese cross-language question answering, Chinese-Vietnamese cross-language information retrieval and other downstream tasks. The experimental results on the MLQA, XQuAD public data sets and the constructed Chinese-Vietnamese cross-language query expansion data set show that the performance indicators of the present application are better than those of the baseline model, and the MAP, Recall, NDCG and MRR are increased by 3.4%, 1.6%, 2.9% and 3.4%, respectively.

A computer-implemented system and method dynamically loads user interface components. A request for a web page is received at a web server coupled to a wide area network from a remote client of a user. An initial user interface for the requested web page, having an associated document object model (DOM), is generated at the web server. An extension registry is initialized and then it is determined that the requested web page includes an extension. The extension registry is queried to identify a storage location for an extension module associated with the extension. The extension module is loaded from the storage location. The extension module implements a user interface component. The extension module is inserted into the DOM. The web page, including the DOM, is provided by the web server to the remote client of the user.

The invention relates to the technical field of artificial intelligence information retrieval, and discloses an intelligent query expansion method and system based on financial knowledge network embedding, and the method comprises the following steps: S1, knowledge network construction and embedding: S11, extracting entities and relationships from multi-source financial data, and constructing a financial knowledge network; and S12, mapping entities and relationships in the financial knowledge network to a vector space by adopting a double-layer embedded framework. According to the intelligent query expansion method based on financial knowledge network embedding, high-quality query expansion is realized by deeply fusing a knowledge network and a semantic understanding technology; the semantic understanding depth of query expansion is improved; through a double-layer knowledge embedding framework and the semantic comprehension capability of a large language model, potential semantics of query can be deeply mined, and the meaning of polysemy words in a specific context can be accurately identified, so that an extension item which better fits the intention of a user is generated, and the knowledge basis of query extension is enriched.

The present application belongs to the technical field of information retrieval, and specifically relates to a retrieval enhancement generation method and system based on knowledge graph and vector retrieval joint scheduling. The present application comprises: receiving a user query text, performing semantic analysis and sub-query expansion using a large language model to generate a multi-path sub-query sequence; based on the sub-query sequence, starting a knowledge graph retrieval channel and a vector knowledge base retrieval channel in parallel to respectively obtain a structured graph document set and an unstructured vector document set; calling a reordering service for relevance reordering on the two document sets respectively; performing intelligent source decision on the two reordering results through a large language model to select an optimal document fusion strategy; assembling the fused context document and the original query into a prompt word, and calling a large language model to generate a final answer and attach provenance information. The present application solves the problem that semantic recall and structured reasoning are difficult to balance under a single retrieval mode, and effectively improves the accuracy and comprehensiveness of the retrieval results.

Systems and methods for performing query expansion. A computing system uses a large language model (LLM) to generate one or more synthetic queries for each document of a set of documents. For a user query, the computing system: selects one or more of the synthetic queries related to the user query; generates an adaptive few-shot prompt to instruct the LLM to generate a response to the query, wherein the adaptive few-shot prompt comprises an example query-response pair for each of the selected one more synthetic queries; provides the adaptive few-shot prompt to the LLM as an input; and generates an amended query based on the output of the LLM in response to the adaptive few-shot prompt.

The invention relates to the field of deep learning, and particularly discloses a deep learning-driven business abnormal behavior prediction method, which comprises the following steps of: firstly, adaptively analyzing an original query, quantifying an intention tendency of the original query, and executing intention-driven heterogeneous query expansion according to the intention tendency; therefore, the subsequent sparse and dense double-channel recall can efficiently cover accurately matched and semantically associated documents at the same time, and the one-sidedness of single-path retrieval is fundamentally overcome. On the basis, a rearrangement module based on a cross encoder is further introduced. The module performs deep interaction and fine-grained correlation scoring on query and candidate documents to realize accurate reordering of candidate lists, and ensures that most critical and most relevant context information is screened and placed at the first place. By means of the mode, the scheme provides high-quality and high-signal-to-noise-ratio input for the final generative large language model, and therefore it is ensured that the generated business anomaly analysis result has high accuracy and high reliability.

The invention relates to an energy hosting method based on retrieval enhancement generation and multi-agent cooperation, and the method comprises the following steps: S1, carrying out the real-time synchronization of a plurality of time sequence operation data of hosting equipment, carrying out the fitting of a plurality of straight line segments and arc segments, and forming a continuous and smooth alternative curve; s2, the geometric description of the alternative curve is combined with the offline document data to be converted into natural language description, vectorization processing is carried out, and a mixed index knowledge base containing a vector database and a knowledge graph is constructed; s3, receiving a user instruction and disassembling the user instruction into a plurality of subtasks; s4, performing query expansion by utilizing concepts related to query entities in the knowledge graph during vector retrieval through the mixed index knowledge base, and reserving a high-score retrieval result as a high-quality knowledge fragment; and S5, generating an energy equipment control strategy by using the high-quality knowledge fragment, the real-time data of the hosting equipment and the security constraint through a large language model.

Various embodiments of the present disclosure provide methods, apparatus, systems, computing devices, computing entities, and / or the like for generating domain-specific queries that are semantically similar to a search query by spell-correcting and tokenizing a search query, and then generating, using an embeddings dictionary data object associated with one or more domain vocabulary data objects, queries semantically related to the search query based on proximity of one or more similar embeddings to an embedding associated with the tokenized query within a domain vector space.

The invention discloses an information reorganization system and method based on a multi-stage thinking chain and retrieval enhancement generation, and the system comprises a multi-source heterogeneous data collection layer which is used for obtaining the structured data and unstructured data of an authoritative data source of a target field in real time, and constructing a field knowledge base; the multi-stage thinking chain processing layer is used for executing thinking chain type question disassembly and query expansion based on user questions, forming a query plan containing a plurality of sub-queries and a dependency relationship thereof, and outputting a structured candidate evidence set after arrangement and retrieval; the retrieval enhancement generation layer is used for obtaining a multi-source retrieval result by adopting a dual-retrieval mode of offline retrieval and online retrieval based on the candidate evidence set; and the visual interaction layer is used for displaying the generated intelligence report and the thinking chain tracing view. According to the method, through innovative combination of a multi-stage thinking processing framework and an RAG technology, the technical problems of data timeliness lag, professional term misalignment and insufficient logic continuity in electric power information processing are effectively solved.

The present application relates to a kind of energy hosting methods based on retrieval enhancement generation and multi-agent cooperation, comprising the following steps: S1, the multiple time series operation data of real-time synchronization hosting device, fitting to obtain several straight line segments and circular arc segment, constitute continuous and smooth alternative curve;S2, the geometric description of the alternative curve is combined with the off-line document material into natural language description, vectorization processing, constructs and constructs the hybrid index knowledge base of vector database and knowledge graph;S3, receive user instruction and be resolved into multiple subtasks;S4, when vector retrieval in the hybrid index knowledge base, use the concept related to query entity in knowledge graph for query expansion, retain the high score retrieval result as high-quality knowledge fragment;S5, the high-quality knowledge fragment and the real-time data of hosting device, security constraint are generated energy equipment control strategy through large language model.

The invention provides a retrieval enhancement generation method and system based on semantic tracing and double query paths, and belongs to the technical field of artificial intelligence. The method comprises the following steps: receiving an original query input by a user, and executing first retrieval to obtain a to-be-remarked text block preliminarily matched with the original query; performing semantic tracing on the to-be-remarked text block, identifying semantic unknown words existing in the to-be-remarked text block, and remarking the semantic unknown words by using an actual reference object to generate a remarked text block; performing semantic extension on the original query for the first time to generate an extended query, and then re-forming a pseudo document according to the query and the extended query; and respectively executing re-retrieval to obtain a first text block set and a second text block set, and finally inputting the first text block set and the second text block set into the large model to generate a final answer. According to the scheme, the completeness of the retrieval context can be ensured, and extra overhead is reduced.

The application discloses a kind of power grid knowledge search query extension method and system, obtain user initial query word, retrieve initial document set from power grid document set;With the preset power grid entity boundary dictionary, the document is cut into text segment, and the context comprehensive weight of each segment is calculated;The semantic similarity of initial query word and each segment is calculated by word vector, and the candidate expansion word in high similarity segment is screened out;Weighted co-occurrence graph is constructed based on the co-occurrence relationship of candidate expansion word, and the context comprehensive weight of common occurrence segment is accumulated to determine the edge weight;Comprehensive sorting is carried out in combination with the word vector similarity with initial query word, and the word in front of ranking is selected as expansion word, and expansion query is formed by merging with initial query word.