36 results about "Markov logic network" patented technology

The invention discloses a deep layer data processing method combined with a knowledge base. The method comprises the following steps that concept tuple sets in a body base are merged; link routes of predicates in different types and corresponding logic rule sets are obtained; the logic rule sets are screened preliminarily, candidate rule sets are obtained; a deep layer probabilistic graphical model is obtained; structuring tuples are obtained based on data to be processed and are mapped to a layering concept space; target tuples are generated, semantic extension is carried out; logic rule sets and evidence tuple sets are obtained; a Markov logical net is subjected to instantiation, the conditional probability for target tuple founding is computed, and data processing results are obtained. The invention further provides a deep layer data processing system which comprises a structuring module, a conceptualization module, a target generating module, an extension module, an activating module and a probability computing module. Context and background knowledge can be fully merged, and accordingly, the purpose of semantic comprehension is really achieved.

The invention provides a news content sensitive word filtering method and system. The method comprises the steps of S1, preprocessing obtained news texts; S2, filtering the sensitive words of the news texts by employing a sensitive word multi-level filtering algorithm on the basis of the priorities of the sensitive words according to a pre-established sensitive word library; S3, judging the filtered sensitive words through an emotion analysis model based on a markov logic network when there are preset sensitive words in the news texts; and S4, marking the news texts as negative news when it is judged that the filtered sensitive words are bad sensitive words, otherwise, marking the news texts as positive news. According to the method and the system, secondary judgment is carried out on the filtered sensitive words through establishment of the emotion analysis model based on the markov logic network, thereby determining whether the filtered sensitive words have negative information or not; therefore, the negative news is filtered; moreover, the positive news fighting against the negative information will not be filtered; and the reliability of filtering the news content sensitive words can be improved.

The invention relates to a system and method for analyzing intelligent behaviors based on scenes and the Markov logic network. The system comprises a scene description machine, an action recognizer and a semantic behavior comprehension describer, wherein the scene description machine is used for classifying the scenes where video images are located according to a theme model method, the action recognizer is used for recognizing atomic action of a person in the video images according to a hidden Markov model method on the basis of the video images, the semantic behavior comprehension describer is used for conducting high-level semantic behavior comprehension and interestingness event description according to a Markov logic network method on the basis of scene classification and atomic action recognition. According to the system and method for analyzing the intelligent behaviors based on the scenes and the Markov logic network, scene description is introduced to high-level semantic behaviors of video to be analyzed, so that the video is more thoroughly described; a field rule knowledge base is introduced to the Markov logic network to be improved, so that high-level semantic behavior description and related event description are achieved more flexibly, and a wider application range is achieved.

The invention discloses an MLN-based network space security situation prediction method and system. The method comprises the steps that asset information data in a specific network space is collected; the collected asset information data is preprocessed, and a network space security situation perception model is constructed and trained; the current network space security situation is evaluated according to the network space security situation perception model and actual data in the current network space; and the future network space security situation is predicted according to a security situation evaluation result of the network space to obtain a security situation prediction result. According to the method, the relation between objects and object properties can be easily represented according to a first-order logic rule by applying a Markov logic network; and in addition, by introducing background knowledge, the object relation in the network space can be mastered more accurately, and then the evaluation accuracy rate is effectively increased. The method and system can be widely applied to the computer network field.

The invention belongs to the technical field of network communication and computers, particularly relates to an intelligent interactive question and answer method, system and device based on a Markovlogic network, and aims to solve the problems that an intelligent question and answer system cannot effectively combine context and background in practical application, cannot feed back in real time and is low in efficiency. The method comprises the following steps: analyzing input information, extracting a structured tuple, and performing semantic expansion by adopting a domain knowledge map; Activating related rules in the domain knowledge map, assigning values to the evidence tuples by adopting an approximate reasoning and / or information input mode, and calculating the posterior probabilityof the candidate response information; And outputting a preset number of pieces of response information with high posterior probability. According to the method, context operation and domain uncertainty knowledge can be effectively fused, approximate deduction and user interaction are powerfully combined, an effective solution is truly provided, meanwhile, automatic induction of knowledge can beachieved, and labor and data cost is reduced.

The invention relates to a Markov logic network-based knowledge mapping relationship type speculation method and device. The device comprises an inference rule obtaining module, a credibility weight learning module, a probability inference module and a relationship type determination module, wherein the inference rule obtaining module is used for generating inference rules according to path features of known nodes of a data knowledge mapping; the credibility weight learning module is used for carrying out credibility weight learning on the inference rules through a Markov logic network and obtaining inference rules with weights; the probability inference module is used for carrying out probability inference on relationship types existing among to-be-speculated nodes according to the inference rules with the weights, so as to obtain relationship type probability among the to-be-speculated nodes; and the relationship type determination module is used for selecting a relationship type with a relatively large probability value as a relationship type among the to-be-speculated nodes according to the relationship type probability obtained by the probability inference module. According to the method and device, the automatic learning of the inference rules in the knowledge mapping and the probability inference of relationship types among nodes can be realized, so that the correctness of speculating the relationship types which possibly exist among the nodes can be effectively ensured.

The invention relates to an association rule transfer learning method based on a Markov logic network. An algorithm of transferring an MLN structure from a source domain to a target domain mainly comprises two parts: firstly, mapping the MLN structure in the source domain to the target domain to establish association between the two domains; and then, optimizing the mapped structure to adapt to indexes of the target domain. The method has the effects that not only can the huge Markov network be concisely described, but also modular knowledge can be flexibly fused into the Markov network. Moreover, imperfection and contradictoriness in a knowledge domain can be tolerated. The algorithm speed is improved. The searching space is restrained by limiting the quantity of updated clauses and updating types of the clauses.

The invention discloses an environment characterization method in intelligent service of a service robot. Knowledge fusion based on vector characteristics is adopted for knowledge from different datasources, a knowledge reasoning algorithm based on a Markov logic network is adopted for further mining the implicit relation existing in the environment information, the autonomous development function of the knowledge base is achieved, and therefore expression work of the indoor environment information based on the knowledge graph is completed. Article relations among environments are stored in agraph form of the knowledge graph, and the search speed of the robot is increased by means of the efficient search capacity of the graph form.

The invention provides a Chinese microblog sentiment classification method and system based on a Markov logic network. The method comprises the steps that microblog data are obtained from the network; predicate and rules used for expressing sentiment classes are determined according to basic knowledge of the markov logic network, corresponding characteristic words are extracted from the microblog data according to a feeling and emotion dictionary, and a MLNs knowledge base is established according to the characteristic words; according to the MLNs knowledge base and predicate and rule training classifiers, sentiment classification is conducted on the microblog data, and Chinese microblog sentiment classification results are output. The method and system can accurately classify sentiments of microblog.

The invention discloses a data association method, system and device based on a Markov logic network. The method includes: utilizing key personnel database, case to be solved and its corresponding weighted rule database to construct the model of obtaining crime probability based on Markov logic network, obtaining the probability of each key personnel committing the case to be solved, thereby screening out the target object, and realizing the data association between the target object and the case to be solved; the method for obtaining the weighted rule base includes: the ontology model view set is constructed by using the data of the solved case which is the same as the main type of the case to be solved, the first-order logic rule set and predicate atom set are extracted, and then the rule weight learning model based on Markov logic network is constructed by using the two extracted sets, the rule weights are obtained by training the model, and thus the weighted rule base composed of the first-order logic rule and the corresponding rule weights is obtained. The acquisition of the rules in the invention does not depend on manpower, and the accuracy rate of data association can be improved.

The invention discloses a complex event identification method based on an ontology model and probability reasoning. The complex event identification method comprises the following steps of (1) modeling a sensor and an event in an intelligent environment by using the ontology model; (2) converting the ontology model into a Markov logic network model by using the semantic attribute of the description logic; (3) segmenting the continuously generated sensor data by using a segmentation method based on a place and a fixed time interval to form an event sequence as the input of a Markov logic network model; (4) carrying out probability reasoning on the Markov logic network model, so that the events occurring in the intelligent environment are identified. According to the complex event recognition method, the advantages of a knowledge driving method and a data driving method are fused, the intelligent environment can be accurately modeled, the time constraint relation between events and the uncertainty of sensing data are effectively processed, and the identification accuracy is improved.

The public opinion role recognition and migration system based on heterogeneous domain migration relates to the fields of data mining and machine learning. In order to solve the problem that the existing technology can not effectively extract knowledge in the face of the complex information of the netizen, can not carry out transfer learning between different fields, and thus can not realize the indirect sharing of knowledge. The system is a public opinion role identification migration model based on the Markov logic network, includes a data predicate module, The structure learning module, theknowledge extraction module, the knowledge transfer module and the parameter learning module. The domain knowledge is converted into the knowledge which can be recognized by the model for structurallearning and extract the knowledge which needs to be transferred to the target domain to complete the knowledge transfer, and then the model after the transfer learning through the parameter learningmodule is obtained. By integrating the conversion complexity into the domain distance and considering the transfer learning boundary from single source domain to single target domain, the migration iseffectively extracted in the face of complex netizen information.

The invention discloses an on-line activity identification method and system based on the Markov logic network. The method includes the steps of acquiring all operation behaviors and storing the operation behaviors in a database to obtain behavior data, processing the behavior data and identifying corresponding activity results according to the behavior data to obtain activity data, and processing the identified activity data and displaying the activity data in a visible way. The system comprises a data acquisition unit, an identification processing unit and a data displaying unit. User activities can be automatically identified from activity logs by establishing an activity identification model, and the expandability is good. Adding logic rules by users is allowed, and strong association rules are mined out from a training sample set by using a rule mining algorithm and serve as logic rules. Uncertainty of corresponding relation between action and on-line activities can be overcome, and the identification accuracy is improved. The method and system can be widely applicable in the computer field.

The invention discloses a text analysis method and device for large-scale multilingual data, and the method comprises the steps: collecting large-scale multilingual text data, and storing the large-scale multilingual text data in a corresponding database; Performing entity matching on the multilingual text data in the database by utilizing a Markov logic network; using the ML-PIB algorithm to carry out clustering analysis on the matched multilingual text data to obtain a target clustering result among the different language information, so that the association contained in the different language information is effectively mined, and the clustering quality is improved.

The invention discloses a smart home decision-making method based on a Markov logic network. The method can be used for solving the problem of uncertainty of intelligent control achieved according touser preferences and environment temperature after a user sends out voice control and natural language processing is conducted, and the uncertainty of the method is due to the fact that evidences provided in the decision making process, namely environment information, are fuzzy. Application of Markov logic network module is the core of the whole method, and the final purpose of realizing decisionmodeling by using an influence graph is realized. According to the method, the weight corresponding to each decision rule is calculated according to parameter learning of the Markov logic network, aninfluence graph model in a specific scene is realized, the maximum expected utility of each decision is calculated, a decision framework in an uncertain intelligent environment is expanded, and the probability of adopting an optimal decision is improved.

The invention comprises a system and article of manufacture to discover potential cyber threats on Twitter. The invention provides a unified framework for modeling and reasoning about the veracity of tweets to discover suspicious users and malicious content. The invention builds on the concept of Markov logic networks (MLNs) for knowledge representation and reasoning under uncertainty.

The invention discloses an automatic data cleaning method based on DeepDive, and the method comprises the steps: (1) comparing an original data scale with a set threshold, and carrying out the randomsampling when the scale exceeds the threshold, so as to obtain sampling data; (2) learning from the original data or the sampling data to obtain a Bayesian network between attributes; (3) converting the Bayesian network obtained by learning into a first-order predicate logic rule; (4) calculating the weight of a first-order predicate logic rule by using a mutual information theory, and convertingthe first-order predicate logic rule with the weight into a Markov logic network; (5) generating a DeepDive rule based on the Markov logic network; (6) carrying out probability reasoning of error / missing data based on DeepDive, and obtaining the probability of taking different values by the attributes of the tuples; And (7) using the reasoning result for cleaning the original dirty data. The method can be used for automatically cleaning the data without existing data quality modes / rules and manual intervention, and the data cleaning efficiency and quality can be effectively improved.

The invention discloses an intelligence use value evaluation method. According to the invention, the indirect action process of the information use value is described by using observation, judgment, decision and attack (OODA) loops; the dynamic uncertainty of the information action process is learned and quantified through a statistical relationship learning framework Markov logic network; therefore, rapid and accurate evaluation of the information use value is realized, and the problem that dynamic and rapid evaluation of existing information use value evaluation is difficult to perform underthe condition of big data is effectively solved.

The invention provides a sensitive word filtering method and device. The method comprises: obtaining a target text; performing word segmentation processing on the target text; performing sensitive word matching on the target text subjected to word segmentation processing, and storing matched sensitive words into a first sensitive word set; determining whether an emotion type to which a context corresponding to the sensitive word in the target text belongs is a specified emotion type or not based on a preset Markov logic network model; deleting the sensitive words corresponding to the context belonging to the specified emotion type from the first sensitive word set to obtain a second sensitive word set; and filtering the target text by using the sensitive words in the second sensitive word set. According to the method, on the premise that the labor cost is reduced, the sensitive word filtering accuracy is improved, and the situation of mistaken killing is avoided.

The invention discloses a travel map-based vehicle destination prediction method, a device, equipment and a medium. The method comprises the steps of obtaining a gate; constructing a traffic map according to the checkpoint data; converting an entity relationship in the traffic travel map into a Markov logic network; mining the traffic travel map to obtain a traffic travel time-space association rule; converting the Markov logic network into an instantiated Markov network; and determining a destination prediction result of the individual vehicle according to the traffic travel time-space association rule and the Markov network. The method can improve the accuracy of individual vehicle travel destination speculation, improves the speculation efficiency, and can be widely applied to the technical field of traffic data processing.

The invention provides an indoor daily activity recognition method in a multi-resident scene. Daily activities in the multi-resident scene can be recognized. The method comprises the steps that a bodymodel of an ADL, a sensor and an environment context is constructed, and user portrayals and an activity sequential relationship are added into the constructed body model, wherein the ADL representsthe daily activities; according to the constructed body model, ADL rules are set up, and a weight value is given to each ADL rule; daily activity data collected by the sensor is acquired, and the daily activities are deduced and recognized according to the daily activity data collected by the sensor, the data collected by the sensor, the constructed body model and the set-up ADL rules based on a Markov logic network. The method is suitable for recognizing the daily activities indoor residents.

The invention comprises a system and article of manufacture to discover potential cyber threats on Twitter. The invention provides a unified framework for modeling and reasoning about the veracity of tweets to discover suspicious users and malicious content. The invention builds on the concept of Markov logic networks (MLNs) for knowledge representation and reasoning under uncertainty.

The invention relates to a Markov logic network-based knowledge mapping relationship type speculation method and device. The device comprises an inference rule obtaining module, a credibility weight learning module, a probability inference module and a relationship type determination module, wherein the inference rule obtaining module is used for generating inference rules according to path features of known nodes of a data knowledge mapping; the credibility weight learning module is used for carrying out credibility weight learning on the inference rules through a Markov logic network and obtaining inference rules with weights; the probability inference module is used for carrying out probability inference on relationship types existing among to-be-speculated nodes according to the inference rules with the weights, so as to obtain relationship type probability among the to-be-speculated nodes; and the relationship type determination module is used for selecting a relationship type with a relatively large probability value as a relationship type among the to-be-speculated nodes according to the relationship type probability obtained by the probability inference module. According to the method and device, the automatic learning of the inference rules in the knowledge mapping and the probability inference of relationship types among nodes can be realized, so that the correctness of speculating the relationship types which possibly exist among the nodes can be effectively ensured.

The invention discloses a deep layer data processing method combined with a knowledge base. The method comprises the following steps that concept tuple sets in a body base are merged; link routes of predicates in different types and corresponding logic rule sets are obtained; the logic rule sets are screened preliminarily, candidate rule sets are obtained; a deep layer probabilistic graphical model is obtained; structuring tuples are obtained based on data to be processed and are mapped to a layering concept space; target tuples are generated, semantic extension is carried out; logic rule sets and evidence tuple sets are obtained; a Markov logical net is subjected to instantiation, the conditional probability for target tuple founding is computed, and data processing results are obtained. The invention further provides a deep layer data processing system which comprises a structuring module, a conceptualization module, a target generating module, an extension module, an activating module and a probability computing module. Context and background knowledge can be fully merged, and accordingly, the purpose of semantic comprehension is really achieved.

The invention discloses an unsupervised data automatic cleaning method, which includes the following steps: A. Data model learning, learning the dependency relationship between attributes from the original data that may contain invalid data, by finding out the implicit non-absolute Or relatively weak dependencies, to obtain a data model expressed in the form of a Bayesian network; B. Generation of data cleaning rules; after obtaining the complete data model of the original data or original data sampling, the data cleaning rules are generated , and specifically generate predicates and first-order predicate rules; C. Generate a Markov logic network based on the predicates and first-order predicate rules generated in step B; D. Generate inference rules based on the Markov logic network generated in step C And data cleaning based on inference results. The method of the invention can effectively improve the data quality of various business systems of the company without consuming a lot of manpower and material resources, and helps the management to make correct decisions.

The invention provides a Chinese microblog sentiment classification method and system based on a Markov logic network. The method comprises the steps that microblog data are obtained from the network; predicate and rules used for expressing sentiment classes are determined according to basic knowledge of the markov logic network, corresponding characteristic words are extracted from the microblog data according to a feeling and emotion dictionary, and a MLNs knowledge base is established according to the characteristic words; according to the MLNs knowledge base and predicate and rule training classifiers, sentiment classification is conducted on the microblog data, and Chinese microblog sentiment classification results are output. The method and system can accurately classify sentiments of microblog.

The present invention provides a method for recognizing indoor daily activities in a scene of multiple occupants, capable of identifying daily activities in a scene of multiple occupants. The method includes: constructing an ontology model of ADL, sensors, and environmental context, and adding user portraits and activity timing relationships into the constructed ontology model, wherein ADL represents daily activities; according to the constructed ontology model, establishing ADL rules, and Assign weights to each ADL rule; obtain the daily activity data collected by the sensor, based on the data collected by the sensor, the constructed ontology model and the established ADL rules, based on the Markov logic network, the reasoning and analysis of daily activities in multi-occupant scenarios can be realized. identify. The invention is suitable for recognizing the daily activities of indoor occupants.

The invention discloses an interpretable link prediction method for a knowledge hypergraph. The method comprises the following steps: constructing an interpretable knowledge hypergraph representation learning model based on a knowledge hypergraph embedding model and a Markov logic network; establishing a joint probability for all observable tuples and hidden tuples of the knowledge hypergraph through a Markov logic network, and maximizing the log likelihood of the observable tuples as a training target; optimizing a confidence lower bound of a log-likelihood function by adopting a variational EM algorithm to realize training and verification of the model; and performing link prediction on the knowledge hypergraph data set by using the verified interpretable knowledge hypergraph representation learning model, namely, taking one hidden tuple in the knowledge hypergraph data set as the input of the model, and outputting a probability value that the hidden tuple is established and the contribution degree of entities and relationships connected with the hidden tuple to the establishment of the hidden tuple by the model. By means of the method, the domain knowledge in the logic rule and semantic information in the vector space can be fully utilized, and the knowledge hypergraph representation learning effect is improved.

The invention discloses an unsupervised data automatic cleaning method. The method comprises the following steps of: A, learning a data model: learning a dependency relationship among attributes fromoriginal data which may contain invalid data, and finding out a hidden non-absolute or relatively weak dependency relationship to obtain a data model represented in a Bayesian network form; B, generating a data cleaning rule; generating a data cleaning rule after obtaining the original data or a complete data model sampled by the original data, and specifically generating a predicate and a first-order predicate rule; c, generating a Markov logic network based on the predicates generated in the step B and the first-order predicate rule; and D, generating an inference rule based on the Markov logic network generated in the step C and cleaning data based on an inference result. According to the method, the data quality of each business system of a company can be effectively improved under the condition that a large amount of manpower and material resources are not consumed, and a management layer is helped to make correct decisions.