CN110334095A - A Method of Association and Search for Human-Machine-Thing Entity Objects Oriented to Trinity Space - Google Patents

Abstract

The present invention relates to a method for associating and searching human-machine-thing entity objects in ternary space, comprising steps: 1) Establishing and expressing cyberspace ternary objects and information collaborative perception and dynamic ternary relationship diagrams: obtaining perception domain objects Collect, extract the data, dimension and feature information of the perceived object, define a new type of logical sensor, as the basic unit of ternary space entity object, information collaborative perception, extraction and expression; 2) create a large-scale ternary relationship diagram and realize the three-dimensional The association of metaspace; 3) Based on the peer-to-peer network architecture, it performs association retrieval for efficient large-scale objects. The beneficial effects of the present invention are: the present invention extracts and retrieves ternary space entities from two aspects of large-scale object collaborative perception and expression and large-scale entity object multi-dimensional association and modeling, abandons fragmented information utilization and integration, Construct the overall connection of ternary space, create an index library for ternary space entities, and perform retrieval operations accurately and efficiently.

Description

A Method of Association and Search for Human-Machine-Thing Entity Objects Oriented to Trinity Space

technical field

The invention relates to the fields of multi-dimensional association technology and large-scale association technology, in particular to an association and search method for human-machine-object entity objects in ternary space.

Background technique

The perception of physical objects in cyberspace includes three aspects: people, machines, and things. It involves wireless sensor networks, Internet information extraction, and crowd sensing related technologies. It has the characteristics of multi-dimensional and large-scale. Among them, the association of cyberspace entity objects involves multi-dimensional association technology and large-scale association technology. Multi-dimensional association technology focuses on the aspect of spatio-temporal data clustering, and divides spatio-temporal objects with similar behaviors into the same group based on spatial and temporal similarity, so that the differences between groups are as large as possible and the differences within groups are as small as possible. Under the large-scale correlation feature, the original clustering algorithm has been optimized. Large-scale entity object search includes semantic search and P2P-based large-scale object search. There are many types of data resources in cyberspace and various forms of expression. Traditional search methods cannot effectively obtain the data resources required by users. Therefore, it is necessary to implement semantic-based intelligent search. In cyberspace, large-scale physical objects must be deployed and stored in a distributed manner. Distributed large-scale entity object search puts forward higher requirements on the implementation technology in terms of search efficiency, scalability, fault tolerance, load balancing and so on. As a typical distributed technology, P2P has a high degree of autonomy, scalability, robustness, and load balancing, which can be used.

In terms of physical object perception, existing solutions obtain information and knowledge from sensor networks, the Internet, and crowd-sensing networks. Fragmentation and scattered distribution require fusion of heterogeneous fragmented information to become a relatively unified and complete information entity.

In terms of entity object association and search, existing technical solutions have begun to focus on the two characteristics of entity objects: multi-dimensional and large-scale. But in cyberspace, the attribute dimensions of entity objects are far more than these. How to realize high-dimensional entity object association and search still has many challenges at the theoretical and technical levels.

Contents of the invention

The purpose of the present invention is to overcome the fragmented information perception and fusion of the existing "human-machine-object" ternary space, and perform information perception acquisition and integration on the multi-dimensional attributes of entity objects. Today's cyberspace covers the Internet, the Internet of Things, and social networks, and contains a large number of physical objects and information in the "ternary space" of people, machines, and things, just like big data. Once these large-scale "ternary objects" are excavated and utilized It will generate huge social value. Therefore, a method for associating and searching human-machine-thing entity objects in ternary space is provided.

This association and search method for human-machine-thing entity objects in ternary space includes the following steps:

1) Establish and express cyberspace ternary objects and information collaborative perception, and dynamic ternary relationship diagrams;

2) Create a large-scale ternary relationship diagram and realize the association of ternary spaces;

3) Based on the peer-to-peer network architecture, it performs association retrieval for efficient large-scale objects.

As a preference: in the shown step 1), establishing and expressing the cyberspace ternary object and information collaborative perception, and the dynamic ternary relationship diagram include the following steps:

1.1) Obtain the collection of perceptual domain objects, extract the data, dimensions and feature information of the perceived objects; fuse the heterogeneous fragment information to form a relatively unified complete information entity; based on the extraction of local object sets, the physical space Adjacent objects complete expressions recognizable by both humans and machines in different information dimensions;

1.2) Define a new type of logical sensor, as the basic unit of ternary space entity object, information collaborative perception, extraction and expression; the internal structure of the logical sensor is divided into two layers according to the function: the first layer is the perception layer, which is used for the three-dimensional space The second layer is the fusion layer, which is responsible for fusing the perceived data, integrating resources and expressing relatively complete object information; the logical sensor performs a three-dimensional space analysis at the two levels of data and resources. information extraction and integration;

1.3) Fusion expression of entity objects: After the unified expression of data and the effective perception of resources are completed at the logical sensor layer, the matching resources and data are fused to form independent entity objects; through technologies such as DNS and URI, entity object resources are realized Unified description of ID, effective addressing and fusion of large-scale entity objects, and efficient retrieval of entity objects on this basis.

As a preference: in the step 2), creating a large-scale ternary relationship diagram and realizing the association of the ternary space includes the following steps:

2.1) Use the basic association layer to identify ternary components: the logic sensor identifies a series of entity objects and extracts information and ID, outputs the basic ternary graph components, and calls the basic association layer to realize the feature extraction of information objects and the ternary relationship graph. complete;

2.2) Use the Peer association layer for large-scale association: perform large-scale association from the perspective of feature values and IDs, and then form a larger-scale entity object association network;

2.3) Use the SuperPeer index layer to form an index library: the SuperPeer index layer further extracts, collects and compresses large-scale feature values and IDs for each large association graph output by the Peer association layer, and finally forms an index library.

As a preference: in the step 3), based on the peer-to-peer network architecture, performing associated retrieval for efficient large-scale objects includes the following steps:

3.1) Feature-oriented intelligent retrieval: build a search problem model based on the user's input according to the semantic model; adopt the method of ontology modeling to perform functional modeling on the user's retrieval requirements;

3.2) Form a feature index system through graph matching, covering, shortest path and other methods; on the basis of the feature index system, a hierarchical fast retrieval method is proposed;

3.3) A new incremental content exchange method is adopted; the incremental information describes the trajectory information of the moving object, and within a given space-time range, the ternary relationship variation is submitted to the adjacent logic sensor through the provided incremental information , and then realize the fast retrieval of moving objects.

The beneficial effects of the present invention are: the present invention extracts and retrieves ternary spatial entities from the two aspects of large-scale object collaborative perception and expression and large-scale entity object multi-dimensional association and modeling, abandons fragmented information utilization and integration, Construct the overall connection of ternary space, create an index library for ternary space entities, and perform retrieval operations accurately and efficiently.

Description of drawings

Fig. 1 is entity object information association and retrieval flowchart of the present invention;

Fig. 2 is a flow chart of the basic entity object cooperative perception and data expression of the present invention;

Fig. 3 is a flowchart of ternary relationship generation of the present invention;

Fig. 4 is a large-scale association and index construction flow chart of the present invention.

Detailed ways

The present invention will be further described below in conjunction with the examples. The description of the following examples is provided only to aid the understanding of the present invention. It should be pointed out that for those skilled in the art, without departing from the principle of the present invention, some improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

The present invention mainly solves the cooperative perception, expression, association and retrieval of ternary space entity objects, and its process is as follows: 1) collaborative perception and acquisition of network space ternary object information; 2) establishment of large-scale ternary relationship diagrams; 3) use of and other network architectures to associate large-scale objects and achieve efficient retrieval.

The present invention is oriented to large-scale entity objects existing in ternary space, involves entity object information perception and association, and finally realizes efficient intelligent retrieval. The basic realization process is as follows: 1) Accurately extract network space entity objects; 2) Large-scale entity objects 3) Based on the peer-to-peer network architecture, efficient and intelligent search for large-scale objects.

The described method for associating and searching for human-machine-thing entity objects in ternary space, the specific steps are as follows:

1. Collaborative perception and expression for large-scale objects:

1) Perception of physical objects:

To obtain the collection of entity objects in the perception domain, it is necessary to consider the characteristics of interweaving and scattered distribution of entity object information in cyberspace. In order to perceive object information in different cyberspaces, different perception methods are required. These perception methods need to work collaboratively and interactively to extract the data, dimensions and feature information of the perceived object.

2) Fusion of isomerized fragment information:

Perceived objects in different cyberspaces are fragmented and heterogeneous. It is necessary to fuse the heterogeneous fragmented information to form a relatively unified and complete information entity, so as to complete the process of extraction and integration from discrete perception to overall objects. According to the validity and continuity of object information in local physical space-time, the integrity, consistency and space-time measurability of each object are further improved.

3) Extract object relationship:

Based on the extraction of local object sets, objects adjacent to the physical space can be expressed in different information dimensions that can be recognized by both humans and machines. Further extract the correlation between objects, and these related human, machine, and object object sets are cooperatively associated with each other to form a "small world" ternary relationship diagram. A large number of such "small world" expressions eventually form an intricate "world" of large-scale entity object associations.

2. Extract cyberspace entity objects:

We define a new type of logical sensor as the basic unit of ternary spatial entity object, information collaborative perception, extraction and expression. The internal structure of the logic sensor is divided into two layers according to functions: the first layer is the perception layer, which perceives the information and resources of human, machine, and object objects in the ternary space; the second layer is the fusion layer, which is responsible for fusing the perceived data, Resource integration and relatively complete object information expression.

1) Perceived entity objects:

① Deploy physical sensors and support efficient communication and information collection:

Firstly, the optimal sensor network deployment in the 2D area is carried out, and the optimal deployment of the ideal sensor circle radius and sector radius and the sensor deployment in bordered areas are discussed. Further realize the sensing of 3D area. Realize the access of physical sensors, network selection, switching, and automatic configuration of network parameters for different network access methods; design an efficient collaborative communication protocol between sensors in the target space. Finally, an efficient, low-energy self-organizing and scalable collaboration and transmission protocol is constructed by using the sensor self-location, initialization self-discovery and neighbor discovery protocols. Real-time collection of environmental monitoring object data. According to the location information, time and sensing objects, the content of each sensor is marked and the connection between the marks is established, as shown in Figure 2.

② Implement and deploy software sensors:

The software sensor is an Internet-based sensing information collection and perception unit, which captures and integrates Internet information. As a new sensor concept, software sensors are deployed differently from physical sensors. It can be deployed locally on a physical sensor or reside on a remote (proxy) server. According to the life cycle, it can be divided into two types: transient and long-term: transient sensors are temporarily defined and flexibly assembled software sensors, which capture a specific information and log off after obtaining the required information; long-term sensors are deployed in network space to Continuous perception of target information.

The input of the software sensor requires Internet information related to the perceived target area, and the output is the basic perception object information and association relationship with a unified perception format, as shown in Figure 2. Due to the wide distribution of network information, none of the existing web crawlers can support the determination of the target area of software sensors and the real-time perception of information. Therefore, real-time perception of object information in the target area is required, and machine learning theories and methods are applied to summarize the perceived information on the Internet, extract feature value IDs, and form basic correlations.

③Swarm intelligence sensor:

Crowd intelligence sensors can perceive the basic information of people and integrate the social and sensory information provided by individuals. The realization of the group intelligence sensor is mainly completed by people's smart terminal APP and the group intelligence sensing server. Realize three basic mechanisms in the crowd-sensing server: 1. Social relationship analysis: apply random graph theory, scale-free network model and other methods to analyze meaningful social relationships. Based on the analysis of large-scale social relations, the analysis of social behavior and form is further carried out. 2. Effective user incentive mechanism: After considering the trade-off between individual information sharing and benefits, a game strategy for maximizing group benefits is formed. 3. Perfect privacy and security guarantee mechanism: formulate detailed security and privacy levels and strict information opening procedures, isolate information according to user levels and information visibility levels, and increase punishment for behaviors that endanger information security, such as reporting , call the police and block contact, etc.

2) Fusion expression entity object:

①Unified expression of perception object data:

The three types of basic sensing units collect data of different dimensions in the ternary space, and the logic sensor realizes the effective perception and integration of object data. Using the object-oriented method, three different types of basic perception data are modeled uniformly, as shown in Figure 2. Among them, the possible sources of standard data objects are three different sensors, but all the specifications are mainly expressed in the form of data records and data labeling relationships.

② Collaborative perception of ubiquitous resources:

First, the logical sensor proposes a standard interface for resource description. Each logical sensor needs to fully understand its own resources and understand the resource information of other logical sensors. Secondly, the extended RDF (xRDF) method is defined to identify and express the relationship between object resources, data and functions sensed by different logical sensors, and on this basis, the unified representation of the ternary relationship graph and its labeling relationship is carried out.

③Effective fusion of entity objects:

After the unified expression of data and the effective perception of resources are completed at the logical sensor layer, the matching resources and data are fused to form independent entity objects. The final entity object builds an effective addressing system. Through DNS, URI and other technologies, the unified description of entity object resource ID is realized, and large-scale entity objects are effectively addressed and fused. On this basis, entity objects are efficiently retrieved. .

3. Effective association and modeling of large-scale entity objects:

The hierarchical modeling method is used to realize the effective association of large-scale entity objects. The specific implementation process is divided into the following three levels, as shown in Figure 4.

1) Use the basic association layer to identify triple components:

First, the logic sensor identifies a series of entity objects, extracts information and ID, outputs the basic ternary graph components, and calls the basic association layer to realize the feature extraction of information objects and the completion of the ternary graph. Using statistics and machine learning methods, the feature value extraction of entity objects is realized, and the nodes of the graph are composed together with the ID. Find the correlation between nodes, form a complete ternary relationship graph, output and submit it to the Peer association layer for the next step of large-scale association. Due to the spatiotemporal correlation and mobility of entity objects, the ternary relationship graph has a dynamic nature, so the variation of the graph can form the trajectory of the moving object.

2) Use the Peer association layer for large-scale association:

The Peer association layer inputs the ternary relationship diagram of the logical sensor output related to the input (see Figure 3), and performs large-scale association from the perspective of feature values and IDs, thereby forming a larger-scale entity object association network. Establish a dynamic associator mechanism in the Peer association layer, that is, dynamically customize the corresponding associator for a specific association relationship. Examples include geographic location correlators, multidimensional attribute correlators, resource correlators, and logical feature correlators. The geographic location correlator is the most intuitive correlator, and the relationship between entity objects in geographic space is the basic relationship. Logical sensors are effectively associated on different attributes to form a multi-dimensional attribute correlator. The method of association uses methods such as progressive matching of large graphs, sparse and dense matrix operations of attributes, and associated intelligent caching to establish the connection between entity objects, and further excavate the association relationship of a large number of ternary relationship graphs. The resource correlator presents the network resources of large-scale logical sensors macroscopically. After superimposing network resources and data resources and strengthening each other, the information association relationship of heterogeneous network space is completed.

3) Use the SuperPeer index layer to form an index library:

Establish a SuperPeer index layer to further extract, collect and compress large-scale feature values and IDs for each large-scale association graph output by the Peer association layer, and finally form an index library. After completing the modeling and design of the Peer association layer, the entity object information is expressed in an explicit or implicit way that can be retrieved. Finally, use the SuperPeer index layer for large-scale fast intelligent retrieval.

4. Efficient and intelligent search for entity objects:

According to the characteristics of large-scale logical sensor networks, an effective index system is established from the perspective of eigenvalues, and efficient and intelligent entity retrieval is realized through mechanisms such as accurate interpretation of user intentions and dynamic processing of moving objects, as shown in Figure 4.

1) Intelligent search:

According to the semantic model, the user's input is used to establish a search question model. The method of ontology modeling is adopted to carry out functional modeling of user's retrieval requirements. Firstly, analyze and identify the functional requirements in the user's retrieval input; secondly, establish a retrieval requirement model through ontology modeling and other methods. At the same time, the analysis and understanding of user search intent is also an important part of intelligent retrieval. Through the analysis of user behavior habits, interest preferences, etc., a search intent model is established for users, and the accuracy of the model is continuously improved through the positive feedback mechanism. , and finally form an intelligent retrieval and analysis tool for target users.

2) Feature index system and hierarchical retrieval:

The semantic model is established starting from the analysis of the user's search intention, which is specifically expressed as the feature set of the entity, and the search result is the information node set that presents the target feature. Therefore, based on the entity object association graph and its network, a feature index system is formed through graph matching, covering, shortest path and other methods. Based on the characteristic index system, a hierarchical fast retrieval method is proposed. In the hierarchical structure, a series of representative nodes are selected in each layer, and then iterative refinement retrieval is carried out sequentially. In the process of retrieval, local sensitivity hash algorithm and feature node selection can be selected to realize.

3) Retrieval optimization of moving objects:

The new incremental content exchange method avoids frequent exchange of content and consumes a lot of resources. Within a given spatio-temporal range, the ternary relationship delta is submitted to the adjacent logical sensors through the provided delta information. Incremental information describes the trajectory information of the moving object, and records and correlates the trajectory information in Peer, thereby realizing the fast retrieval of the moving object.

Claims (4)

1. a kind of association and searching method towards ternary space people's-machine-object entity object, which is characterized in that including following step It is rapid:

1) establish and express three meta object of cyberspace and information synergism perception, dynamic ternary relation graph；

2) it creates extensive ternary relation graph and realizes the association in ternary space；

3) it is based on grid architecture, is associated retrieval for efficient extensive object.

2. the association and searching method according to claim 1 towards ternary space people's-machine-object entity object, feature It is, in shown step 1), establishes and express three meta object of cyberspace and information synergism perception, dynamic ternary relation graph include Following steps:

1.1) perception field object set is obtained, data, dimension and characteristic information are extracted, it is real to form relatively uniform complete information Body；Then the object adjacent to physical space completes people and all identifiable expression of machine in different information dimensions；

1.2) novel logical sensor is defined, as ternary spatial entity, the base for information synergism perception, extracting and expressing This unit；The internal structure of logical sensor is divided into two layers by function: first layer is sensing layer, in ternary space people, machine, Object object information and resource are perceived；The second layer is fused layer, is responsible for the data of fusion perception, reallocates resources and express phase To complete object information；Logical sensor on two levels of data and resource to the information in ternary space extract with it is whole It closes；

1.3) amalgamation and expression entity object: after completing the Unified Expression of data and the effectively perceive of resource in logical sensor layer, The resource to match and data are merged, independent entity object is formed；By DNS, URI technology, entity object is realized The Unify legislation of resource ID, to extensive entity object carry out desired address with merge, progress entity object on this basis Efficient retrieval.

3. the association and searching method according to claim 1 towards ternary space people's-machine-object entity object, feature Be, in the step 2), create extensive ternary relation graph and realize the association in ternary space the following steps are included:

2.1) identify ternary component using basic associated layers: logical sensor identifies a series of entity object and extracts information And ID, basic ternary diagram component is exported, basic associated layers is called to realize feature extraction and the ternary relation graph of information object It is complete；

2.2) it is associated with: being associated on a large scale in characteristic value with the angle of ID, in turn on a large scale using Peer associated layers Form more massive entity object related network；

2.3) form index database using SuperPeer index level: SuperPeer index level exports Peer associated layers each big Associated diagram carries out extraction, collection and the compression of further extensive characteristic value and ID, ultimately forms index database.

4. the association and searching method according to claim 1 towards ternary space people's-machine-object entity object, feature It is, in the step 3), is based on grid architecture, being associated retrieval for efficient extensive object includes following step It is rapid:

3.1) input of user the intelligent retrieval of Feature Oriented: is established into search problem model according to semantic model；Take ontology The method of modeling carries out functionalization modeling to the Search Requirement of user；

3.2) aspect indexing system is formed by the matching of figure, covering, shortest path, then proposes with different levels quick-searching side Method；

3.3) new increment type content exchange method is used；Increment information describes the trace information of mobile object, what is given In space-time unique, ternary relation variable quantity is submitted to by the increment information of offer by adjacent logical sensor, and then realize To the quick-searching of mobile object.