264results about How to "Implement mining" patented technology

The invention relates to a Tibetan language entity knowledge information extraction method, which comprises the following steps that: Tibetan and Chinese comparable language material information is extracted from Tibetan and Chinese text language material information; entity equivalence pairs are extracted from the Tibetan and Chinese comparable language material information; the Tibetan and Chinese cross-language entity relationship is extracted from the entity equivalence pairs; a Tibetan language "entity-attribute-value" triad is extracted from the Tibetan and Chinese cross-language entity relationship; and the triad is stored into a Tibetan language entity knowledge semantic resource library. The Tibetan language entity knowledge information extraction method solves the problem of Tibetan language training language material deficiency to a certain degree, promotes the knowledge sharing among different languages, and provides support for the study in the fields of Tibetan and Chinese cross-language knowledge questions, information retrieval, machine translation and the like.

The invention relates to the fields of deep learning, natural language processing and the like, and particularly relates to a syntactic relationship enhanced machine reading understanding multi-hop reasoning model and method. The model comprises a text coding module, an association element relationship graph construction module, a multi-hop reasoning module, an answer generation module and an answer prediction module. According to the invention, syntactic relationships are fused into the graph construction process, an associated element relationship graph is constructed, multi-hop reasoning iscarried out by utilizing a graph attention network based on the relationship graph, and answer support sentences are mined; meanwhile, a multi-head self-attention mechanism is introduced to further mine text clues of viewpoint type questions in the article, and an automatic solution method of the viewpoint type questions is improved; and finally, a plurality of tasks are subjected to joint optimization learning, so that when the model answers the questions, fact description for supporting the answers can be given, the interpretability of the model is improved, and meanwhile, the existing answering method for viewpoint type questions is improved.

The invention discloses a wireless network protocol knowledge graph construction analysis method and system, equipment and a medium, and the method comprises the steps: defining entities, the relation between the entities and a triad according to the endogenous factors of a wireless communication network protocol, and constructing a knowledge graph of the endogenous factors of the wireless communication network protocol through the triad; obtaining sparse representation vectors of the nodes through node random state weight calculation, using cosine similarity among the sparse representation vectors for quantifying and replacing edge connection relations in the knowledge graph, and then calculating association vectors and feature vectors of the nodes. According to the method, the knowledge graph of the wireless communication network protocol endogenous factors is established, a node sparse representation algorithm based on a knowledge graph structure and a cosine similarity calculation and node association analysis method are introduced, and graph structure information contained in the knowledge graph is deeply mined; quantitative feature representation is carried out on the knowledge graph node and edge connection relation, and analysis and mining of the incidence relation between the nodes are completed.

The invention provides a core content mining method and equipment for large-scale voice data and belongs to the technical field of computers. According to the core content mining method and equipmentfor the large-scale voice data, which are provided by the embodiments of the invention, a to-be-processed voice data set can be converted into a corresponding to-be-processed text data set, and then text clustering is carried out on a to-be-processed word text set corresponding to the to-be-processed text data set through a text clustering algorithm in order to obtain at least one corresponding category, a subject corresponding to the at least one category is determined as a core content of the to-be-processed voice data set, and a subject of the category where each piece of to-be-processed text data in the to-be-processed text data set belongs is determined as the core content of the to-be-processed text data, namely the core content of each piece of to-be-processed voice data is determined. According to the method and the equipment, mining of the core content of the large-scale voice data is realized under the zero priori condition, and the efficiency and the accuracy rate of miningof the core content are improved.

The invention provides a method for comprehensively analyzing and processing real-time alarms on the basis of attack strategy graphs and an intrusion detection system thereof; splitting of attack scene graphs can be effectively prevented, the complete attack scene graphs can be reconstructed, the subsequent attack can be predicted, attack scenes can be fused with one another, and alarm information capable of being directly utilized can be accurately provided for analysts. The method for comprehensively analyzing and processing the real-time alarm includes steps of (1), constructing the attack strategy graphs; (2), reconstructing attack scene graphs, adding missing alarms or inferences of attack links omitted deliberately by an attacker, utilizing inference results as inference alarms and adding the inference alarms into alarm sets to be associated; (3), predicting subsequent attacks; (4), fusing the attack scene graphs, expressing fused associated records by graphs, and acquiring fresh attack scene graphs. The method and the intrusion detection system have the advantages that splitting of attack scene graphs can be effectively prevented, and analyzing accuracy can be improved. The intrusion detection system is used for analyzing and processing safety events to alarms generated by the system by the analyzing and processing method.

The invention discloses a word forming determination model generation method, and a new word discovery method and device. The new word discovery method includes: performing pre-processing on a text so as to extract a plurality of text blocks; acquiring the word frequency, the cohesion degree and the coupling degree of each text block as word forming feature information of each text block; using a pre-generated word forming determination model and the word forming feature information to classify each text block so as to recognize new words. Automatic discovery of the new words can be achieved, the word forming feature information of each block includes the word frequency, the cohesion degree, and the coupling degree; and the accuracy of new word recognition can be improved.

The invention discloses an intelligent perception and prediction method and device for a network space situation and computer equipment, and relates to the technical field of network security. The method comprises the following steps: extracting and expressing network space entity concepts and relationships and attributes between the entity concepts, constructing a cyberspace entity concept set, associating and fusing multi-source heterogeneous original data with the cyberspace entity concept set, constructing a cyberspace situation knowledge graph, and finally performing multi-target behaviordynamic evolution analysis and prediction according to the cyberspace situation knowledge graph. Deep information mining of network space multi-source heterogeneous data is achieved, and then comprehensive and deep analysis and prediction of the overall operation situation of the network are achieved.

The invention relates to a method for video semantic mining, comprising the steps of: firstly performing a Chinese continuous speech recognition, a video target recognition and a video character recognition on a to-be-processed video; then performing a Chinese word division and a part-of-speech tagging on the recognition result, reserving nouns and verbs as the peaks of a graph model, wherein a side weight between the peaks is set to be the Chinese semantic distance of words represented by the two peaks; and finally mining the semantic information of the video according to a dense subgraph finding algorithm. The semantic mining of the video is realized by the fusion of the three recognition results of the Chinese continuous speech recognition, the video target recognition and the video character recognition; the video is represented to be a graph model, wherein the peaks are the words in the video, and the side weight is set to be the semantic distance between the two peaks; the video semantic mining algorithm is further transformed to be the dense subgraph finding algorithm of the graph model. The method and the device in the invention solve the problems of high error rate of a single recognition result and incapability of efficiently fusing a plurality of recognition results in the process of the Chinese continuous speech recognition, the video target recognition and the video character recognition, as well as solve the problem of the video structured expression and the problem of the video semantic mining algorithm realization. The method and the device in the invention can be used for performing automatic marking, classification and semantic mining on batches of videos.

The invention provides a construction method and construction device for a data warehouse model. The method comprises the steps of constructing a business model by business modeling and conducting abstraction processing on the business model to generate a domain model; dividing the domain model into a plurality of topic models by using multiple topics; determining the mapping relationship betweenentities of each topic model according to the business logic; constructing the data warehouse model according to the mapping relationship between the entities of each topic model. The method divides the domain model into multiple topic models, which is beneficial for fully and effectively achieving data analysis and mining; the mapping relationship between the entities of each topic model is determined according to the business logic, the data warehouse model is constructed, and the stability and scalability of data are effectively improved.

The embodiment of the invention discloses a method for discriminating abnormal data, which is realized on a plurality of sample data sets. The method comprises the following steps: acquiring a first sample data set and a plurality of corresponding types of service; setting screening rules in each corresponding type of service and obtaining screened data of each type of service according to the set screening rules; judging whether the screened data of each type of service exist in screened comparison data sets of the plurality of sample data sets except the first sample data set or not; if so, determining the screened data as abnormal data. According to the embodiment of the invention, the problem that analysis results are greatly different due to deviation existing in the sampling process can be corrected, the method can be used for sampling of a plurality of subtypes and the error rate of sampling results is reduced; meanwhile, all abnormal data in complex big data (data aggregate instead of sample sets only) can be quickly and accurately locked.

The invention discloses a multi-dimensional electric energy meter on-site state test method based on a decision-making tree mass. The method comprises the following steps of building a decision-makingtree by using collected data related to an electric energy meter and provided by an electricity information collection system; forming next-level branches through classifying each decision-making tree; generating the completely grown primary decision-making tree; pruning and optimizing on the primary decision-making tree based on a cost complexity pruning algorithm; using an independent test setto assess the accuracy of the primary decision-making tree after pruning and optimizing; selecting the optimal decision-making tree from each set of secondary decision-making trees, and forming a decision-making tree abnormity diagnostic model by all the optimal decision-making trees, wherein a judgment result wins in a voting manner; and transmitting an operational rule of the decision-making tree abnormity diagnostic model to an electricity information collection system Hadoop big data processing cluster, thus implementing abnormity detection on the abnormal running state of the running electric energy meter.

The invention relates to an intelligent four-quadrant flood regulation method for middle-sized and small-sized reservoirs. According to the intelligent four-quadrant flood regulation method, the rainfall-runoff relation P+W0-R or W, the peak discharge relation R or W-Qm, the flood regulation relation V-Q, the relation between the total primary flood amount, the rise reservoir capacity and the maximum reservoir water storage W-V<qi>-Vm are drawn in a four-quadrant diagram relating to centering; the runoff depth R or the total primary flood amount W corresponding to the precipitation P is obtained according to the peak discharge relation R or W-Qm curve; the Qm corresponding to the runoff depth R or the total primary flood amount W is obtained according to the peak discharge relation R or W-Qm curve; and the Vm corresponding to the runoff depth R or the total primary flood amount W is obtained according to the relation curve between the total primary flood amount, the rise reservoir capacity and the maximum reservoir water storage W-V<qi>-Vm; and the intersection point of a straight line between the Qm and the Vm and the flood regulation V-Q curve is obtained, and the reservoir capacity V and the spillway discharging capacity Q afteer flood regulation are obtained. According to the intelligent four-quadrant flood regulation method, the relevant curves of the reservoir inflow prediction and reservoir flood routing calculation of the reservoirs are drawn on the diagram, the setting time of a flood regulation scheme can be shortened, and the prediction timeliness of the small-sized reservoirs is effectively improved.

The invention provides an analysis method and device for multi-mode radiomics, and a terminal. The method comprises steps of obtaining multiple mode images, and preprocessing the multiple mode images; performing regional segmentation to the preprocessed mode images to obtain an area of interest corresponding to each mode image; for each area of interest of each mode image, performing high-pass feature extraction, and obtaining characteristics corresponding to each area of interest; grouping the characteristics corresponding to each area of interest of the multiple mode images into source characteristics, and using a preset clustering algorithm to perform characteristic clustering of the source characteristics; and according to the result of the characteristic clustering, constructing radiomics markers. Therefore, the problem that image characteristics cannot be extracted from multiple aspects in the radiomics research in the prior art is solved, the number of the source characteristics for the characteristic clustering and the representative characteristic kind after clustering are enriched, and the medical image information can be excavated to the greatest extent.

The invention relates to a public opinion analysis method, which comprises the steps of collecting public opinion data including project publicity data and public opinions under the project publicitydata, reporting the project publicity data by media, and discussing the project publicity data by the public; preprocessing the public opinion data to obtain structured public opinion information; generating a public opinion object according to the public opinion information, wherein the public opinion object comprises an object identifier, an object category, space-time information, semantic information, emotion information and relationship information; conducting similarity matching the public opinion objects and public opinion cases in a public opinion case library to acquire the most similar public opinion cases; and analyzing the most similar public opinion cases to obtain a public opinion control scheme. The network public opinion data can be reasonably mined, massive multi-source public opinion data can be quickly processed, valuable public opinion information can be obtained, and data support is provided for subsequent public opinion case similarity matching and public opinioncontrol scheme analysis.

The invention discloses a meteorological sensitive electric quantity mining method capable of considering multi-region difference. The method comprehensively considers the non-linear relationship between the electric quantities of a plurality of mined regions and various meteorological factors such as temperature, humidity and rainfall; the long-term trend component, namely, amount of natural increase of the electric quantities can be finely stripped by technologies such as X-12-ARIMA, so that the change rule of the electric quantities along with weather conditions and day types is effectively acquired, and precise mining of daily meteorological sensitive electric quantities at the specific regions can be realized; the meteorological sensitive electric quantities can be used for better carrying out deep analysis on the root cause of electric quantity change, so that short-term electric quantity forecasting is guided.

The invention discloses a draft interaction based monitoring video visual analysis method. The method comprises the steps of 1) marking a monitoring video and adding characteristic attribute description information for a set target in a target case; and then extracting key frames from the video based on the mark information and building a key frame index; 2) drawing a related draft of the target case; and carrying out related marking on a node connecting line of the related draft and carrying out node information marking around the node of the related draft; 3) identifying the structure of the related draft and calculating a directed and weighed graph of the related draft; 4) carrying out related analysis on the directed and weighed graph, obtaining a key path of the related draft and obtaining a closely related event set of the target case; and 5) calculating the similarity of the related draft sub-graph of the target case, the related draft sub-graph structure and other case related drafts to obtain similar sub-graphs of the related draft sub-graph of the target case. The draft interaction based monitoring video visual analysis method can efficiently describe user intentions and enhance video semantics.

The invention discloses an electronic commerce service platform, which comprises a corporate customer management system, a transaction platform, an online negotiation system, an online payment system, a data acquisition system and a commerce short message system, wherein the corporate customer management system comprises a customer relationship management (CRM) system, a corporate post office, a supply chain management system and an office automation (OA) system; the data acquisition system comprises an information acquisition module, a query and statistical module, a data analysis module, an information processing and flow control module, a user and right management module and a system maintenance module. The electronic commerce service platform is an integrated corporate electronic commerce tool capable of providing self-service website establishment, office automation, a small customer relationship management system, the supply chain system, a corporate mail system, short messages, online negotiation transaction, mobile equipment retrieval and browse and the like for the corporate.

The invention discloses an active voice detection method based on noise scene recognition, and the method comprises the steps: extracting an optimal feature facing a noise classification task from anaudio signal, inputting a feature value into a noise type classifier, so as to recognize a noise type in the audio signal, according to the noise type, determining a preferred feature and a classifiersuitable for the voice and noise classification task, extracting optimal features oriented to the voice and noise classification task from the audio signals, inputting the optimal features into a voice and noise classifier, and judging whether the voice signals exist in the audio signals or not. According to activity voice detection system based on noise scene recognition, before binary classification of noisy voice and noise signals is carried out, according to the method, the current noise type is detected and identified, the most distinctive feature combination is preferably selected for the specific noise type, model parameters can be designed for the specific noise type, and the effectiveness and stability of the performance of the whole detection process under different noise typesare ensured.

The invention discloses a multi-modal parameter model optimization fusion method based on image omics features. The method comprises the following steps: obtaining medical images of a plurality of modes, and preprocessing the medical images; performing region segmentation on the preprocessed medical images to obtain a region of interest corresponding to each modal medical image; obtaining a high-dimensional imaging omics feature corresponding to each region of interest; dividing the high-dimensional image omics features to obtain a training set and a test set, and performing gradient dimensionality reduction to obtain low-dimensional image omics feature tags; and training a plurality of candidate parameter models by using the low-dimensional imaging omics feature tags as a cross training data set and using different algorithms. A subject working feature curve is drawn, the area mean value under the subject working feature curve is calculated, and the area mean value with the highest area mean value is determined as the optimal parameter model, so that the image features can be effectively extracted, and medical image information can be mined to the maximum extent.