16 results about "Web retrieval" patented technology

A computerized method for generating an answer to an input question, the method has the steps of: identifying, in a knowledge graph (KG), one or more entities included in the input question; finding, in the KG, a plurality of reasoning paths for the one or more entities based on comparison of a similarity between an embedding of each of the plurality of reasoning paths and an embedding of the input question; retrieving a plurality of quotes from a network, each quote being a piece of text related to the input question taken verbatim from the network; and sending the input question, and an input set to a foundation model (FM) to obtain the answer, the input set comprising the plurality of reasoning paths and the plurality of quotes.

The invention discloses a reinforcement learning system and method fusing multi-source deep retrieval and reasoning ability, and belongs to the technical field of artificial intelligence. The system comprises a large-model-driven reasoning agent which works in cooperation with a multi-source toolkit integrating local retrieval and network retrieval, and a large-model-driven reasoning agent which works in cooperation with the multi-source toolkit integrating local retrieval and network retrieval. Through a reinforcement learning framework designed by the invention, when the intelligent agent is trained to solve a complex problem, not only can the time of information retrieval be autonomously decided, but also which retrieval tool including a local knowledge base or network search can be intelligently selected and called, and a retrieval result is seamlessly fused into a multi-step reasoning chain. According to the method, the accuracy, timeliness and reliability of an artificial intelligence model in tasks such as question answering and reasoning are effectively improved, and the problems of knowledge illusion and static property of an existing large model are solved.

The invention discloses an automatic express parcel sorting system based on artificial intelligence, which relates to the technical field of automatic express parcel sorting and comprises a multi-mode sensing layer unit, an intelligent control module, a normal sorting execution mechanism, an abnormal sorting execution mechanism, an abnormal supplement identification module and an abnormal compensation processing device. The multi-mode sensing layer unit integrates double information acquisition modes of visual sensing and label scanning, and can effectively deal with complex scenes such as label abrasion, fouling and folding. And the abnormal sorting execution mechanism and the abnormal supplementary recognition module realize automatic distribution and multi-dimensional supplementary recognition of abnormal packages. Multiple sets of visual supplementary sensing modules and label supplementary scanning modules are used for omnibearing acquisition of six surfaces of the parcel, so that the problem of blind area identification of a limited area on the surface of the parcel by a traditional system is solved; and the network retrieval module carries out information completion through a compliance logistics information platform interface to form an automatic processing chain for identifying, shunting and supplementing identification information completion, so that the manual intervention demand is greatly reduced.

The invention provides a large model event analysis enhancement method based on multi-dimensional retrieval enhancement generation, and relates to the technical field of artificial intelligence and information analysis crossing. After to-be-retrieved information is obtained, on one hand, a network retrieval result is inquired on a network, an image-text consistency evaluation result and a semantic authenticity evaluation result are obtained according to the network retrieval result, and first corpus data are obtained by integrating the network retrieval result, the image-text consistency evaluation result and the semantic authenticity evaluation result; on the other hand, the to-be-retrieved information generates retrieval vectors in multiple dimensions, a knowledge retrieval result of each dimension is obtained through retrieval in a corresponding preset knowledge database, and the knowledge retrieval results of multiple dimensions are integrated to obtain second corpus data. And combining the first corpus data and the second corpus data to obtain an enhanced corpus, and taking the enhanced corpus as semantic data to be analyzed by a large model. When a large model executes an analysis task by using the semantic data provided by the scheme of the invention, an analysis result with higher credibility can be obtained.

Techniques are described for providing multifactor authentication using randomly selected Digital Identity factors associated with a human user or individual seeking access to a facility, building, computer system, network, application, memory, etc. Digital Identity factors are or memory. Digital Identity factors are retrieved and stored by an authentication agent. The Digital Identity factors can be retrieved from a network such as the internet or can be previously provided by a customer or human individual requesting authorization. Upon receiving a request for authentication, a plurality of Digital Authentication factors are randomly selected. A request is sent to an environment associated with the human individual seeking access, the request including information regarding the randomly selected Digital Authentication factors. A reply is received having information regarding the randomly selected authentication factors. The information in the reply is compared with the randomly Digital Identity factors to determine whether to grant or deny access.

The application discloses a video time retrieval method based on feature representation enhancement, and steps are as follows: generating a training set; constructing a video time retrieval network; training the video time retrieval network; and retrieving a video time. The application constructs a video time retrieval network based on feature representation enhancement, adds a query timing correction operation to a query feature extraction module, arranges all subject-predicate-object triples describing different actions in a to-be-queried sentence according to time sequences of the corresponding actions, realizes cross-modal feature sequence order matching, and reduces cross-modal semantic alignment difficulty. The application constructs a double-connection graph convolution module, takes candidate segment features as graph nodes, connects the candidate segment with global candidate segments of the front and rear parts through graph convolution, realizes fusion of the candidate segment features and the video global features, and improves candidate segment feature representation quality.

The invention discloses a text interaction method, system and device based on a large language model and a medium, and belongs to the technical field of artificial intelligence. The method comprises the steps that a query statement is received, a query vector is obtained according to the query statement, and at least one local vector and a corresponding local text segment are obtained in a vector database according to the query vector; performing network retrieval according to the search keyword list of the query statement to obtain a network retrieval result, segmenting the network retrieval result to obtain at least one network text segment, and vectorizing the at least one network text segment to obtain a corresponding paragraph vector; determining a target text segment based on the sorting result; inputting the query statement and the target text segment into a large language model to obtain text interaction feedback information corresponding to the query statement; through multi-source retrieval and correlation coefficient reordering, the expansion of the retrieval coverage rate is realized, and the timeliness of the retrieval result is improved.

The invention discloses a network retrieval enhancement generation method and system oriented to large model cue word optimization, and belongs to the field of natural language processing. The method comprises the steps of obtaining an instruction input by a user; performing word segmentation on the instruction by using a word segmentation tool and extracting keywords; performing network retrieval through an external search engine according to the keywords to obtain a retrieval document set; performing text segmentation processing on the retrieval document set to obtain a plurality of text blocks; vectorizing the text blocks and storing the vectorized text blocks in a vector database; retrieving text blocks related to the instruction from a vector database to obtain a related text block set; reordering the related text block set and selecting text blocks as enhanced contexts; integrating the enhanced context with the user instruction to generate a final cue word; and inputting the final cue word into the large model, and obtaining a response result output by the large model. According to the invention, the answer accuracy, correlation and practicability of the LLM in diversified scenes are improved.

The invention discloses a joint optimization retrieval enhancement generation system and method, and relates to the technical field of artificial intelligence, and the system comprises a strategy network, a retriever and a reinforcement learning module; wherein the strategy network is used for determining whether retrieval is needed or not according to an input state, generating a retrieval query and generating a final answer; the retriever is used for returning related information according to a retrieval query generated by the strategy network; and the reinforcement learning module is used for performing joint optimization on the strategy network and the retriever through a reinforcement learning algorithm. The strategy network and the retriever are subjected to collaborative optimization through the reinforcement learning algorithm, so that the system can integrally optimize collaborative work among all the modules, the overall performance of the system is remarkably improved, retrieval efficiency, answer generation quality, response speed and the like are remarkably improved, and by dynamically adjusting behaviors of all the modules, the search efficiency of the system is improved. The system can be excellent in performance in different input states, and the requirement of a user for efficient and accurate knowledge acquisition is met.

A method is disclosed. A data set including: (a) identifiers of a set of incidents occurring within a defined geographic region to which at least one service provider responded during a first time period and (b) address data identifying a location within the geographic region of each said incident of the set is retrieved over a network. An instruction to generate a heat map of the incidents occurring within the geographic region during the first time period is received from a user via a user interface generated to a display device. In response to the instruction to generate the heat map, the address data is converted to GPS data. A heat map of an aerial view of the geographic region based on the GPS data is generated. The heat map is displayed to the display device in a user interface.

This invention relates to the field of sustainability, and particularly to the fields of sustainable production, end-product reuse, and end-of-life recycling. This disclosure relates to methods, apparatus, and computer elements for retrieving data from a distributed network associated with output products and / or input materials used to produce such output products.

An attribution management system for rewarding view-based attribution on an online platform includes an attribution server that retrieves from a business manager over a data communication network using an Application Programming Interface (API), a set of candidate posts. Pixel traffic from the set of candidate posts is tracked based on first party data and social media posts. User interaction, including clicks and views, is identified on the set of candidate posts. Candidate posts below a pre-determined threshold value are boosted using a predetermined number of user interactions. A source of the one or more boosted candidate posts is identified using tags. At least one purchase from the views on one or more boosted candidate posts is identified. An attribution amount for each owner of the one or more boosted candidate posts is determined based on the at least one purchase using the source identified by the tags.

Embedded resource layouts are presented by rendering an interface document to cause a graphical interface to be displayed in accordance with appearance specifications for the graphical interface in the interface document, embedding, into the interface document in response to rendering the interface document, a resource layout document corresponding to a reference in the interface document and a user identity, retrieving, through a network in response to embedding the resource layout document, a plurality of resource values, each resource value among the plurality of resource values corresponding to a resource reference in the resource layout document, and arranging the plurality of resource values in the graphical interface in accordance with positioning specifications in the resource layout document.

This invention discloses a multimodal agent-enhanced retrieval algorithm, belonging to the field of enhanced retrieval algorithm technology. The invention includes a multimodal text conversion unit, a vector database construction unit, and a multimodal knowledge graph (MMKG) construction unit. It enhances cross-modal reasoning capabilities, solving the modality disconnect problem in existing technologies: establishing visual-text associations through MMKG. It improves the efficiency of multi-source knowledge synthesis by employing plug-in parallel retrieval. Through consistency verification of the decision-making agent and expert refinement, it reduces the answer illusion rate, i.e., the proportion that conflicts with facts; MMKG embeds visual data storage paths, and web retrieval provides URL tracing, allowing users to trace the source of each fact in the answer, thus solving the black-box reasoning problem in existing RAGs.

The invention provides an interview data processing method and device based on a large model, electronic equipment, a storage medium and a computer program product, relates to the technical field of artificial intelligence, in particular to the technical fields of large models, natural language understanding, simulated interview and the like, and can be applied to simulated interview scenes. According to the specific implementation scheme, a retrieval request is generated through a first large model according to question and answer data in the interview process; performing network retrieval according to the retrieval request to obtain reference data; and through the second large model, generating answer optimization data of the question and answer data according to the reference data. According to the method, the artificial intelligence large model and the network retrieval are combined, so that the professionality and the accuracy of the answer optimization data are improved, the interview skill of the interviewer is improved, and the simulated interview requirement of the interviewer is met.

A computerized method for generating an answer to an input question, the method has the steps of: identifying, in a knowledge graph (KG), one or more entities included in the input question; finding, in the KG, a plurality of reasoning paths for the one or more entities based on comparison of a similarity between an embedding of each of the plurality of reasoning paths and an embedding of the input question; retrieving a plurality of quotes from a network, each quote being a piece of text related to the input question taken verbatim from the network; and sending the input question, and an input set to a foundation model (FM) to obtain the answer, the input set comprising the plurality of reasoning paths and the plurality of quotes.