38 results about "Event evolution" patented technology

The invention discloses a news event evolution analysis method based on time sequence distribution information and a topic model and relates to the field of text analysis. The method comprises the following steps: firstly, dividing a corpus into a plurality of sub-corpuses according to time by analyzing distribution characteristics, presented on a time sequence, of a news report, and by using a K-Means clustering algorithm; secondly, sequentially performing topic modeling on each sub-corpus by using the topic model, and learning the model through a Gibbs sampling method to obtain topic distribution information of each sub-corpus; finally, connecting topics between which the distance is minimum in series by calculating a Jensen-Shannon distance between each two topics in the adjacent sub-corpuses, wherein the topics are connected in series to obtain a main topic of an event, and auxiliary topics except the main topic in each sub-corpus are concerns and new developments of the event in each stage. According to the method, the mainline of event development in a news prediction and new concerns burst in each stage can be better described.

The invention relates to an event evolution knowledge ontology-based event prediction method and apparatus, and a terminal device. The method comprises the following steps: 1, , an event evolution ontology knowledge base with an event evolution relationship is constructed; extracting a first event chain and a first event tuple from an event evolution ontology knowledge base, remote supervision mode, utilizing the first event tuple and the first event chain to perform relational data remarking on the unstructured text; according to the method for establishing the event prediction model throughevent data training, a large amount of marked training data can be effectively and quickly obtained based on knowledge, and therefore the accuracy of event prediction is effectively improved.

The invention discloses a method and device for generating an event group evolution diagram and solves the technical problem of high time complexity in operating an algorithm for configuring event evolution diagrams. The method includes: acquiring one event evolution diagram which represents corresponding relationship between duration and attention hotness of events; acquiring an available area unoccupied by other event evolution diagrams from candidate areas corresponding to the duration according to a segment tree; configuring the event evolution diagram in the available area so as to generate the event group evolution diagram. By the method and device, time complexity in operating the algorithm for configuring the event evolution diagrams is lowered.

The invention relates to a method for studying and judging financial market changes based on a fact map and a multi-factor model. The method comprises the steps that S1, acquiring various historical events with causal relations in a historical event library, carrying out aggregation evolution according to the influence relations among the historical events to constructthe fact map; s2, matching the obtained current event as a result event with historical events in a fact map, and extracting various upstream historical events having an influence relationship on the current event to generate anevent evolution chain; s3, extracting reason factors of each historical event directly or indirectly influencing the current event on the event evolution chain based on the factor library; s4, selecting conventional factors in the cause factors to construct a cause factor set influencing the current event change, and calculating the explanation degree of the cause factor set on the current resultfactor; and if the interpretation degree is lower than the interpretation degree threshold, introducing an abnormal factor in the cause factor into the cause factor set until the interpretation degreethreshold is met. The method is high in studying and judging efficiency and high in accuracy.

The embodiment of the invention provides a method and device for generating an evolution diagram of an event group. The method comprises the steps: obtaining a plurality of event evolution diagrams which indicate the evolution state of the popularity of concerned events; sequentially marking the drawing positions of a plurality of events in a combination candidate region according to the time sections of the plurality of events and the popularity of the corresponding events; and respectively enabling the plurality of event evolution diagrams to be copied in the combination candidate region according to the drawing positions, so as to generate the evolution diagram of the event group. When the position of the evolution diagram of any one event in the combined drawing region is determined, there is no need of considering the layout of other events, and the longitudinal drawing position just needs to be determined according to longitudinal heights corresponding to marked time points, thereby reducing the time complexity in layout of the event evolution diagrams, and reducing the time complexity from O(nlgn) to O(n).

The invention relates to an interpretable financial subject matter generation method and device fusing emotion, industrial chain and cause logic. The method comprises the following steps: establishinga cause-and-effect logic library by utilizing a cause-and-effect relationship between events; performing event identification and event linking on the input event based on a cause logic library to obtain a cause and effect event evolution chain; associating the causal event evolution chain with an industry chain knowledge base, and obtaining an entity industry chain conduction chain through association reasoning of upstream and downstream of the entity; performing attribute-level sentiment analysis on entities in the entity industrial chain conduction chain; and performing influence reasoningon the financial subject matter according to the reason logic library, the industrial chain knowledge base and the emotion analysis result, and generating the financial subject matter and an associated reasoning result of the financial subject matter. According to the method and device, reasoning is carried out by fully utilizing the rule of combining reason evolution logic and industrial chain reasoning logic, the dynamics, evolutionary property and conductivity of the reasoning process are fully reflected, and the reasoning result is more credible.

The invention discloses a visual analysis method and system for the dynamic event evolution of social media. The method comprises the following steps that: aiming at certain event to extract a keywordfrom the social media, and obtaining social media data from the social media according to the keyword; according to a processing rule based on map metaphor, adopting a map mapping way to convert theinput social media data into an event evolution map E-Map, wherein the processing rule based on the map metaphor comprises cities, towns and rivers; and forming a structured projection space which contains semantics, and exploring the projection space through interaction by a user, wherein the social media data send to the processing rule at least comprises the following data: keywords, releasinginformation and retransmission information containing the keywords, and a user information spread behavior. By use of the method and the system, events in the social media can be subjected to visual analysis, and visualization can provide a high-level conclusion for event evolution for user, and also can enable the user to own ability for deeply exploring different aspects of data.

The invention discloses a social network event evolution tracking method based on time division, is applied to the technical field of information, and aims to solve the problem of tracking errors of an evolution tracking method in the prior art. The method comprises the steps that firstly, collecting text information in a user publishing dynamic state from all social networking sites, and conducting event feature extraction on the text information; secondly, obtaining a plurality of event evolution relationships according to the obtained events, and generating an event evolution graph by adopting a time division method; then selecting a starting event and a terminating event according to the focus event of interest; and finally, solving an optimal path to each event by using a viterbi algorithm. Compared with an existing method, the event evolution tracking effect is better.

The embodiment of the invention provides an analysis method and an analysis system based on an event evolution process of key words. The method comprises the steps of counting the quantity of search results within all time quanta, so as to obtain a sequence that the quantity of the search results changes along with the time quanta; performing peak detection on the sequence, so as to obtain at least one peak window; performing text analysis on the search result corresponding to the peak window, so as to obtain peak description; displaying the sequence that the quantity of the search results changes along with the time quanta, and displaying the peak description in a peak position. According to the analysis method and the analysis system of the embodiment of the invention, the sequence that the quantity of the search results changes along with the time quanta is counted, and the sequence that the quantity of the search results changes along with the time quanta and the peak description are displayed, so that the trend that the microblog information quantity changes along with time is shown, the evolution process of an event related to the key words can be shown by the trend that the microblog information quantity changes along with time, a user can know about main contents of microblog information without reading microblog contents in detail according to the peak description, and the information quantity displayed by the search results is increased.

The invention discloses an event discovery and display method based on a knowledge graph. The method comprises the following steps: S100, preprocessing each piece of government service hotline data; s200, dividing the government service hotline data in the time period selected by the user into data frames according to a time interval specified by the user, and performing event detection on the government service hotline data in each frame through a community discovery algorithm to obtain each city event; and S300, based on a bipartite graph maximum weight matching algorithm, matching events of front and back frames, establishing an event chain of each city event evolution, analyzing z-score values of each city event in the event chain, determining emergencies and emergency degrees of each city event, and performing early warning and multi-dimensional visualization on the city emergencies therein. According to the invention, various events occurring in a city can be clearly known, and detection and early warning can be carried out on emergencies in the city.

The invention belongs to the technical field of artificial intelligence, and particularly relates to a narrative event prediction method fusing event environment information. The method includes the steps that firstly, environment description directly related to a narrative event is obtained through the internal relation between the narrative event and a non-structural text; secondly, local and global event environment characteristics are extracted from environment description by utilizing semantic information of events through a multi-head attention mechanism; thirdly, mutual coding is performed on the narrative event characteristics and the environment characteristics through an event prediction model coding module to obtain event evolution characteristics of a fusion event environment; and finally, a narrative event prediction decoding module selects a correct next narrative event from the candidate items according to the evolution characteristics, and an event prediction task is completed. According to the method, the problem of insufficient utilization of event environment information in a narrative event prediction task can be solved, the event environment information can provide more perfect context information for a narrative event, and the understanding level of an event prediction model for event evolution is improved.

The invention discloses an event chain-based network public opinion association deduction prediction analysis method, which comprises the following steps: extracting related node events from microblog article text data according to event development, and forming a theme evolution tree and a public opinion evolution probability graph through clustering association and mode mining matching prediction, wherein corpora of a microblog training set are pre-trained through an ELMO model, and microblog articles are vectorized; combining the similar microblog data through a One-Pass clustering algorithm to obtain a node event set; performing few-label data learning through ActiveLearning, and the label quality is improved through man-machine interaction; and discovering a transfer logarithm between tags through Seq2Pat, and further constructing a Markov chain, forming a topic tree and predicting a node event evolution trend. According to the invention, extraction, association and prediction of sub-events in complex public opinion events are made up, a man-machine interaction label method with expert intervention is fully considered, and the invention is suitable for public opinion deduction analysis positioning and accurate prediction.

The invention relates to an intelligent deduction method for a large-area power failure event based on a Bayesian network, which comprises the steps of summarizing the general evolution process of power security emergencies through typical large-area power failure cases; analyzing multiple paths of large-area power failure event evolution by surrounding three elements such as the scenario state, processing method and processing target, selecting key scenarios, and constructing a scenario network of event development; calculating the conditional probability of hopping between network nodes by combining historical cases and expert opinions; and building a Bayesian network of the large-area power failure event by means of simulation, and realizing the multi-path and multi-result deduction. The intelligent deduction method effectively solves the problems of uncertainty and multiple paths of scenario deduction of the typical power security emergencies, can visually reflect the effect of emergency measures, and facilitates the emergency personnel to timely adjust disposal measures and continuously improve the emergency level. The intelligent deduction method has important significance for analyzing the evolution mechanism of emergencies.

The invention discloses a big data-based emergency event evolution reasoning method and system. The method comprises the steps of integrating cases in a database to obtain a sub-event chain set; taking the sub-event chain set as an input of an algorithm; obtaining weak and strong rule item sets through the algorithm; obtaining the strong rule item set from sets higher than the support degree, and obtaining the weak rule item set from sets lower than or equal to the support degree, wherein the strong rule item set serves as a conventional emergency event evolution possible path, and the weak rule item set serves as a non-conventional emergency event evolution possible path; and by taking the weak and strong rule item sets as reasoning bases, mining a subsequent sub-event set with a maximum occurrence probability and a subsequent sub-event set with an extremely small occurrence probability from the case database. Therefore, the deduction that an Apriori algorithm only aims at the strong rule item set can be improved, and the weak and strong rule item sets in a calculation process are distinguished and both reserved as target data.

The invention relates to an event evolution knowledge graph construction method, device and equipment, and a storage medium. The method comprises the steps of obtaining a Chinese news report of news;extracting corresponding event elements from the Chinese news report, and combining the event elements into a Chinese event corresponding to the Chinese news report, the event elements comprising event trigger words and event arguments; based on event arguments in the event elements, aiming at each piece of different news, associating a plurality of different Chinese events corresponding to each piece of different news; and according to the priori cause relationship of the event trigger words, constructing an event evolution knowledge graph among a plurality of associated different Chinese events. According to the method, the problem of low accuracy of the evolution event in the prior art can be relieved, and the technical effect of improving the accuracy of the evolution event is achieved.

The embodiment of the invention provides an event evolution relationship analysis method and device, and the method comprises the steps: respectively carrying out the extraction of event elements of two inputted events, and generating respective event element sets of the two events; and based on the event element set, obtaining evolution characteristics used for analyzing an evolution relationshipbetween the two events, calculating a deviation degree of each evolution characteristic along with time change, and obtaining an evolution relationship analysis result of the two events. According tothe event evolution relationship analysis method and device, the evolution relationship of the event is analyzed by using event elements such as entities and keywords of the news event, the dynamic relation between the events is obtained, the problem that an existing analysis method only describes the static relation in the events is solved, a user can rapidly master global information and evolved veins of the events, and the method has important significance in judgment of the development trend of the events.

The invention discloses a cascading failure emergency blocking method based on an accident chain model. The method comprises the following steps: firstly offline computing and searching a high-risk accident chain by using a electromechanical transient time domain simulation method based on actual power grid data, screening out an accident chain set with first two accidents unconfigured with emergency control measure from the high-risk accident chain by considering the existing emergency control system configuration information; secondly, configuring the emergency control policy for the secondaccident of this class accident chain by adopting the offline computing method, and ensuring that the cascading event evolution process can be blocked; and finally, determining the continuous definitehappening of the first and the second accidents in the accident chain as the necessary condition for triggering the emergency control measure on the basis of accomplishing the real-time matching of the operation way by the emergency control system, wherein the economy and the timeliness of the control are both taken into consideration. Through the method disclosed by the invention, the cascadingfailure caused by the continuous disturbance development unconfigured with the emergency control measure can be timely blocked, the miss of the control opportunity is avoided, and the system loss is reduced.

The invention discloses a microblog event evolution analysis method based on a feature word co-occurrence graph. The method comprises the steps: carrying out the data preprocessing of an original event microblog set, and generating an event microblog word set; extracting feature words of the event microblog word set, and constructing a feature word co-occurrence graph based on a feature word co-occurrence relationship in the text; carrying out feature word community division on the feature word co-occurrence graph; and calculating the similarity between the microblogs and the feature words inthe community, and carrying out sub-event microblog clustering processing. According to the invention, a user can be helped to quickly and fully understand the evolution process of the whole popular event, so that interested information can be quickly and accurately obtained.

The invention provides a method for immediately tracking random event evolution based on Internet mass information. According to the method, an initial spatial vector of an event is constructed by using basic characteristic elements, in the tracking process, entropy values of frequently-used words of Internet increment information are computed and counted on the basis of the initial spatial vector, the growth and structural change of the initial spatial vector of the event are determined by taking the entropy values as event evolution power, and the stable fission of new event space is realized by the entropy value comparison of an increment information vector and an event critical space vector, so that the identification and tracking of a derivative / secondary event are realized. By the method, the entropy values of the increment information are taken as the power of the event evolution, so that the requirements of work such as tracking, pre-warning / predicting emergency events and random events can be effectively met; and meanwhile, the method is high in topicality, reliability and expandability, and the common phenomenon that the load of computational nodes and a network is heavy in a mass information processing activity is avoided.

The embodiment of the invention provides an atlas construction method, an early warning method, equipment and a medium, which are used for accurately constructing an affair atlas. In the embodiment of the invention, the knowledge graph of each industrial chain is created for the industrial chain, and after the knowledge graph of each industrial chain is created, the created knowledge graphs are connected to form the intermediate knowledge graph, and the incidence relation between the intermediate knowledge graph and the event in the event evolution rule is accurately established, therefore, the event atlas can be accurately established in the embodiment of the invention.

The embodiment of the invention provides an analysis method and an analysis system based on an event evolution process of key words. The method comprises the steps of counting the quantity of search results within all time quanta, so as to obtain a sequence that the quantity of the search results changes along with the time quanta; performing peak detection on the sequence, so as to obtain at least one peak window; performing text analysis on the search result corresponding to the peak window, so as to obtain peak description; displaying the sequence that the quantity of the search results changes along with the time quanta, and displaying the peak description in a peak position. According to the analysis method and the analysis system of the embodiment of the invention, the sequence that the quantity of the search results changes along with the time quanta is counted, and the sequence that the quantity of the search results changes along with the time quanta and the peak description are displayed, so that the trend that the microblog information quantity changes along with time is shown, the evolution process of an event related to the key words can be shown by the trend that the microblog information quantity changes along with time, a user can know about main contents of microblog information without reading microblog contents in detail according to the peak description, and the information quantity displayed by the search results is increased.

The invention discloses an analysis method for event evolution process integration in a physical world and a network world. The method comprises the following steps: (1) carrying out data integration;defining an analysis target; extracting a dynamic evolution process of an event in the physical world; clearing text data in the network world, such as advertisements, declarations and the like; extracting entities, keywords and popularity of the physical world and the network world; extracting emotional numerical characteristics of the user response; designing a cubic data structure organization, and supporting the visual display and data query tasks of users; (2) carrying out visual integration; designing a proper visualization scheme based on basic visual structures such as straight lines,curves and the like; marking symbols, intersection points and connecting lines of the physical world events; performing visual display on emotional tendency and discussion keywords generated by social media users to related events of the entity; (3) carrying out interactive integration; subjecting the interaction part and the visualization part to multi-level conversion; and supporting application-oriented composite interaction deep analysis is supported.

The invention discloses a method for deducing an event evolutionary graph path. The method comprises the following steps: S1, presetting an event evolutionary relation template; S2, obtaining various types of source data from the data source according to an event detection algorithm; S3, analyzing the obtained source data; S4, preprocessing the analyzed source data, and performing event extraction to form a data network; S5, carrying out prediction in combination with the data network, and obtaining prediction information; S6, establishing an event evolutionary graph according to a preset event evolutionary relation template and the prediction information; and S7, summarizing the event evolutionary relation events to generate an event evolutionary event library. According to the method, the analyzed source data is preprocessed, events are extracted, the data network is formed, and the corresponding event evolutionary graph is established, so that the integrity and continuity of a generated event evolution process can be effectively improved, and the complexity of deducing an event evolutionary graph path is reduced.

The invention relates to an improved script learning method and device based on an event evolution diagram. The method comprises the following steps: constructing an abstract event evolution diagram by utilizing a specific event network formed by specific event chains; mining event evolution knowledge contained in the event evolution diagram to learn initial embedded representations of context events and candidate events; inputting the initial embedding representations of the context event and the candidate event into a bidirectional LSTM network to obtain updated embedding representations of the context event and the candidate event, including time sequence information of an event chain and event evolution knowledge contained in an event evolution diagram; and carrying out similarity calculation on the context events and the candidate events by utilizing the updated embedded representation to obtain a similarity score of each candidate event, probabilizing the similarity scores, and selecting the candidate event with the highest probability as a final predicted event. The method has high accuracy for script event prediction, and can be used in the fields of privacy stealing attack event prediction and the like.