CN102855272B - A kind of micro-blog contains the D-S evidence theory method of traffic information fusion - Google Patents

Abstract

The invention discloses a D-S evidence theory method for microblog containing traffic information fusion. The method includes: determining the identification framework and proposition space of the evidence theory according to the degree of traffic congestion; The microblog content related to the information theme is composed of the microblog data set that needs to be integrated; the microblog data set is preprocessed; the word meaning similarity of the traffic state description vocabulary of the microblog data set is calculated, and the Chinese corpus resources are introduced to improve the calculation accuracy of the word meaning similarity; Calculate the weighted evaluation of microblog word meaning similarity, construct the basic probability distribution function of evidence theory; perform evidence synthesis on multiple microblog evidence sources according to the Dempster combination rule, determine the trust interval of each proposition in the identification space, and select the trust interval with the largest trust function propositions as a fusion result. The invention can realize the integration of traffic information contained in microblogs, and provide an important data source for the collection of urban traffic information.

Description

A D-S Evidence Theory Method of Microblog Containing Traffic Information Fusion

technical field

The invention relates to mobile location service, Internet space information search, and mobile Internet technology, in particular to a D-S evidence theory method for fusion of traffic information contained in microblogs.

Background technique

Real-time traffic information can alleviate traffic congestion, improve transportation efficiency, ensure traffic safety, and facilitate public travel. The existing methods of obtaining traffic information mainly include fixed sensor technology (induction coils, video surveillance, and microwave detection), and the installation of GPS and wireless communication equipment. However, these collection methods still have great limitations in obtaining temporary traffic control restriction information and responding to sudden traffic incidents. Microblogs contain a wealth of real-time traffic information with high timeliness, covering various types of traffic information, such as road traffic flow, road smoothness and driving speed, traffic restrictions, temporary traffic control, sudden traffic incidents, The traffic status description information of a specific location, etc., and the acquisition of highly dynamic real-time traffic information contained in microblogs can make up for the shortcomings of existing traffic information collection methods.

However, the high dynamics and ambiguity of microblog messages and the differences in the descriptions of messages posted by different microblog users make information fusion a bottleneck in information extraction, which directly affects the utilization of real-time traffic information contained in microblogs. Microblog traffic information fusion is to reason and make decisions on traffic information content contained in different microblog messages, obtain accurate description information of traffic status, and better serve traffic management and travel services. The difficulty of microblog traffic information fusion lies in: (1) The unstructured characteristics of microblog messages cause semantic understanding difficulties: because the content of microblog messages is simplified, only about 140 characters, and the characteristics of colloquial language are obvious, it contains more redundant content. It puts a lot of pressure on automatic semantic judgment and extraction; (2) Different microblog users post messages that describe traffic conditions differently, causing information collection contradictions: within a certain period of time, there may be multiple microblogs describing the traffic conditions of the same road segment information. For the same road condition, the descriptions of different users may be very different, and some descriptions are even semantically mutually exclusive.

To solve this problem, the technology currently used is the text clustering method. When the text has a certain number of words, the text clustering process can determine the accurate text topic description. However, the content of microblog messages is short, and after the process of word segmentation, word meaning disambiguation, and word meaning drying, there are very few key words that can be used to describe traffic conditions. Therefore, text clustering cannot well solve the ambiguity of microblog message descriptions and the differences in message descriptions of different microblog users.

Therefore, this patent proposes a microblog traffic information fusion method based on the D-S evidence theory for the above information fusion problem. This method enriches the semantic information of microblogs by introducing Chinese corpus resources, and solves the fuzzy problem of microblog descriptions. Based on the knowledge of the Chinese corpus, the weighted evaluation of word meaning similarity to the microblog content is realized, and then the evidence theory is used to deal with the uncertainty reasoning problem of information fusion caused by the differences of different microblog users, so as to determine the fusion result . It provides a new solution for dynamic traffic information collection and makes up for the relatively weak link of traditional dynamic traffic information collection technology.

Contents of the invention

The technical problem to be solved by the present invention is: it is difficult to fully integrate and utilize a large amount of traffic information contained in the current microblog, and it is difficult for the traditional dynamic traffic information collection method to respond to the current situation of sudden road condition information in a timely manner. The invention proposes a D-S evidence theory method of microblog containing traffic information fusion, which provides another important data source for the collection of dynamic traffic information. This method solves the difficulty in semantic understanding caused by the unstructured features of microblog messages and the contradictions in information collection caused by the differences in the description of traffic status in the messages posted by different microblog users. It can be directly applied to personal and vehicle navigation, mobile location services, map websites, Professional travel information service platform, logistics dispatch and traffic emergency plan.

Technical solution of the present invention is:

A D-S evidence theory method for the fusion of traffic information in microblog, including:

According to the degree of traffic congestion, determine the evidence theory identification framework Θ and the proposition space 2 ^Θ ;

According to the effective time window T _interval and the road network section road, capture the microblog content related to the topic of traffic information, and form the microblog data set V that needs to be fused;

The microblog information V _i preprocessing operation includes natural language word segmentation, word meaning disambiguation, and word meaning drying to obtain the microblog traffic state description vocabulary set W _i ;

Introduce the Chinese corpus resource Corpus={C _wikipedia , C _hownet ,..., C _num }, and calculate the word sense similarity Sim between the microblog traffic state vocabulary set W _i and the words in the proposition space;

Calculate the word meaning similarity weighted evaluation Score _i of the microblog message V _i , and determine the basic probability distribution function m(V _i ) of the evidence theory;

Carry out evidence synthesis and evidence decision-making through the Dempster synthesis rule, and determine that the road microblog of this road section contains the traffic information fusion result TState _road ;

Preferably, the identification framework for determining the evidence theory according to the degree of traffic congestion includes:

The determination of traffic congestion can refer to some national standards, such as the "Urban Traffic Management Evaluation Index System" published by the Ministry of Public Security in 2002 to classify road network traffic congestion;

The determination of the degree of traffic congestion is related to the actual integration needs;

There is no limit to the size of evidence theory identification framework and proposition space;

Preferably, according to the effective time window and the road network section, grabbing the relevant microblogs of traffic information topics, forming the microblog data set V that needs to be fused includes:

The effective time window T is defined as the time period formed by expanding the traffic information time T _current , that is, T=[T _current -Δt _a , T _current +Δt _b ] where Δt _a and Δt _b are user-defined parameters;

The road section in the road network takes the road name of the road network as the research object or takes the road segment in the navigation road network as the research object;

The microblog data set V is related to the actual selected road network section;

The construction process of the microblog dataset V is not involved in this patent;

Microblog dataset V does not limit its storage form, it can be a database or a data file;

Preferably, the introduction of the Chinese corpus resource Corpus to calculate the word sense similarity Sim between the microblog traffic state vocabulary set Wi and the vocabulary in the proposition space includes:

The Chinese corpus resource Corpus does not limit the type of corpus, it can be Wikipedia, HowNet, etc.;

The calculation process of word meaning similarity Sim is not involved in this patent;

Preferably, the word meaning similarity similarity weighted evaluation Score of described micro-blog content, the basic probability distribution function m (V _i ) that determines micro-blog comprises:

The weighted evaluation calculation of word meaning similarity in microblog content is as follows:

${\mathrm{<span\; class="notranslate">\; Score</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; T</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; Sim</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; term</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; log</span>}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; num</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; sum</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; boost</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; user</span>}<span\; class="notranslate">\; )</span>$

Among them, k is a proposition in the proposition space of evidence theory, term microblog traffic state description vocabulary set W _i vocabulary, sum is the number of microblogs, num(k) is the number of microblogs containing proposition k, boost(user) is the The incentive function of a microblog user reflects the importance of the user. The default value is 1. The larger the value, the more important the user is.

The calculation of the basic probability distribution function of microblog content is as follows:

$\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; Score</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; Score</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}$

Preferably, the method also includes:

According to the desired fusion road section and fusion time conditions provided by the user, complete the fusion of traffic information contained in the microblog;

According to the road section and fusion time condition in the desired fusion area provided by the user, complete the fusion of traffic information contained in the microblog;

Compared with the prior art, the present invention has the advantages that: the present invention overcomes the shortcoming that the traffic information contained in the micro-blog is difficult to be fully integrated and utilized; by introducing Chinese corpus resources, the semantic understanding process of the micro-blog is enriched, and the unstructured information of the micro-blog is effectively solved. Difficulties in understanding semantics due to culturalized features; using the advantages of D-S evidence theory in dealing with the expression and synthesis of uncertain information, it solves the contradictions in information collection caused by the differences in the descriptions of traffic states in the messages posted by different microblog users.

Description of drawings

Fig. 1 is the flow chart of the theoretical method of D-S evidence fusion of traffic information contained in the microblog of the embodiment of the present invention;

Fig. 2 is the flow chart of the implementation of the case of the embodiment of the present invention

Fig. 3 is the case microblog information record collection of the embodiment of the present invention

Fig. 4 is the Wikipedia data source of the case of the embodiment of the present invention

detailed description

In order to enable those skilled in the art to better understand the solutions of the embodiments of the present invention, the embodiments of the present invention will be further described in detail below in conjunction with the drawings and implementations.

As shown in Figure 1, it is a flow chart of a D-S evidence theory method for the fusion of traffic information contained in a microblog according to an embodiment of the present invention, including the following steps:

Step 101, according to the degree of traffic congestion, determine evidence theory identification framework Θ and proposition space 2 ^Θ ;

Step 102, according to the effective time window T _interval and the road network section road, capture the microblog content related to the traffic information topic, and form the microblog data set V that needs to be fused;

Step 103, the microblog information V _i preprocessing operation, including natural language word segmentation, word meaning disambiguation, and word meaning drying, to obtain the microblog traffic state description vocabulary set W _i ;

Step 104, introducing the Chinese corpus resource Corpus={C _wikipedia , C _hownet , ..., C _num }, calculating the word sense similarity Sim between the microblog traffic state vocabulary set W _i and the vocabulary in the proposition space;

Step 105, calculating the word meaning similarity weighted evaluation Score _i of the microblog message V _i , and determining the basic probability distribution function m(V _i ) of the evidence theory;

Step 106, carry out evidence synthesis and evidence decision-making through Dempster synthesis rule, and determine that the road section road microblog contains traffic information fusion result TState _road ;

The actual application process of the embodiment of the present invention will be further described in detail below by taking the road network of Beijing as an example.

As shown in FIG. 2 , it is a flow chart of a case implementation of a D-S evidence theory method embodiment of a microblog containing traffic information fusion in the present invention. This embodiment is implemented under the premise of this embodiment, and detailed implementation and specific operation process are provided, but the scope of protection of the present invention is not limited to the following embodiments, and the specific process of this embodiment includes the following step:

Step 201, referring to the regulations in the "Urban Traffic Management Evaluation System" promulgated by the Ministry of Public Security of my country in 2002, its traffic congestion mainly includes the following four types: smooth (the average speed of motor vehicles is greater than 30km/h), mild congestion ( The average speed of motor vehicles is 20-30km/h), congestion (the average speed of motor vehicles is 10-20km/h), congestion (the average speed of motor vehicles is less than 10km/h). Since the microblog content is an overview description of traffic conditions, it is not difficult to be accurate to the speed level. In order to ensure the accuracy of microblog information fusion, mild congestion and congestion are combined, and the identification framework is defined as;

Θ = {smooth, congested, congested}

Then the proposition space 2 ^Θ is:

Step 202, loop through all road sections in the Beijing road network, and select and process the Middle East Fifth Ring Road section of the road network;

Step 203, judging whether all road sections in the road network have been processed, if yes, execute step 217; if not, execute step 204;

Step 204, constructing a time window, and establishing a microblog information record related to traffic information topics related to the East Fifth Ring Road by searching microblog content such as Sina microblog, Netease microblog, Sohu microblog, and Tencent microblog about dynamic traffic information in Beijing set, as shown in Figure 3;

Step 205, microblog information preprocessing, natural language word segmentation, word meaning disambiguation, word meaning drying, to obtain microblog traffic state description vocabulary, and then perform step 206;

Step 206, download and analyze the Wikipedia data source (as shown in Figure 4), using the literature (Milne D, Witten IH.An effective low-cost measure of semantic relatedness obtained from Wikipedia links.In Proceedings of the AAAI 2008 Workshop on Wikipedia and Artificial Intelligence) method sets up semantic model, calculates the semantic similarity of micro-blog traffic state vocabulary, then executes step 207;

Step 207, calculate the weighted evaluation of the word meaning similarity of the microblog message, determine the basic probability distribution function of the evidence theory, then execute step 208;

Step 208, carry out evidence synthesis and evidence decision-making through Dempster synthesis rule, and determine that the microblog of this road section contains traffic information fusion results, then execute step 209;

Step 209, judge whether all road sections in Beijing have been processed, if yes, execute step 217; if not, execute step 210;

Step 210, end the calculation process;

It can be seen that in the embodiment of the present invention, a D-S evidence theory method for the fusion of traffic information contained in microblogs can solve the problem of semantic understanding difficulties in the process of fusion of traffic information contained in microblogs and the problem of description differences between multiple microblog users, providing dynamic traffic information. Information collection provides a new information collection scheme to accurately and quickly respond to unexpected road condition information. The invention can be directly applied to release of dynamic traffic information, and serves map website system, public travel information platform and mobile location service.

It should be noted that the method of the embodiment of the present invention is applicable to the dynamic information collection of all urban road networks; the present invention does not limit the grabbing method of microblog content and the website providing microblog content; the present invention is not limited to the specific implementation The word-sense similarity calculation method used and the weighted evaluation model of word-sense similarity for Weibo users.

Those of ordinary skill in the art can understand that all or part of the steps in the method of the above-mentioned embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage Media, such as: ROM/RAM, disk, CD, etc.

The embodiments of the present invention have been described in detail above, and the present invention has been described by using specific embodiments herein. The description of the above embodiments is only used to help understand the method of the present invention; meanwhile, for those of ordinary skill in the art, according to this The idea of the invention will have changes in the specific implementation and scope of application. To sum up, the contents of this specification should not be construed as limiting the present invention.

Claims (6)

1. a micro-blog contains the D-S evidence theory method of traffic information fusion, it is characterised in that step includes:

According to congested in traffic degree, conclusion evidence theory framework of identification Θ and proposition space 2^Θ；

According to effective time window T_intervalWith road network section road, capturing the micro-blog content relevant to transport information theme, composition needs the micro-blog data set V merged；

Micro-blog message V_iPretreatment operation, including natural language participle, word sense disambiguation, meaning of a word mummification, obtains micro-blog traffic behavior word finder W_i；

Introduce Chinese corpus resource Corpus={C_wikipedia, C_hownet... }, calculate micro-blog traffic behavior word finder W_iWith the acceptation similarity Sim of vocabulary in proposition space；

Calculate micro-blog message V_iAcceptation similarity weighting evaluation Score_i, conclusion evidence theory Basic probability assignment function m (V_i)；

Carry out combining evidences and evidence decision by Dempster compositional rule, determine that this section road micro-blog contains traffic information fusion result TState_road。

Method the most according to claim 1, it is characterised in that described include according to congested in traffic degree conclusion evidence theory framework of identification:

Road network traffic congestion degree is classified by " urban traffic control assessment indicator system " that the determination of congested in traffic degree is announced with reference to the Ministry of Public Security for 2002；

The determination of congested in traffic degree is relevant to actual fused demand；

Evidence theory framework of identification and proposition space are not intended to size.

Method the most according to claim 1, it is characterised in that described according to effective time window and road network section, captures the micro-blog content relevant to transport information theme, and composition needs to merge micro-blog data set V and includes:

Effective time window T_intervalIt is defined as transport information time T_currentThe time period carrying out expanding and formed, i.e. T_interval=[T_current-Δt_a, T_current+Δt_b] wherein Δ t_aWith Δ t_bFor user's defined parameters；

In road network, section is with road network road name as object of study or with the roadway segment in navigation road network as object of study；

Micro-blog data set V is relevant to the actual road network section chosen；

Micro-blog data set V is not intended to its storage form.

Method the most according to claim 1, it is characterised in that described to introducing Chinese corpus resource Corpus, calculates micro-blog traffic behavior word finder W_iInclude with the acceptation similarity Sim of vocabulary in proposition space:

Chinese corpus resource Corpus is not intended to corpus type.

Method the most according to claim 1, it is characterised in that the acceptation similarity weighting evaluation Score of described micro-blog content, determines the Basic probability assignment function m (V of micro-blog_i) including:

Micro-blog content meaning of a word Similarity-Weighted evaluation calculation, particularly as follows:

Wherein, k is the proposition in evidence policy innovation space, and term micro-blog traffic behavior descriptor collects W_iVocabulary, sum is the quantity of micro-blog, and num (k) is the micro-blog quantity comprising proposition k, boost (user) is the excitation function of this micro-blog user, reflecting the significance level of this user, default value is 1, and this user of the biggest explanation is the most important for this value；

The Basic probability assignment function of micro-blog content calculates, particularly as follows:

6. according to the method described in any one of claim 1 to 5, it is characterised in that described method also includes:

The section to be merged provided according to user and time of fusion condition, complete described micro-blog and contain traffic information fusion；

According to section and time of fusion condition in the wanted integration region that user provides, complete described micro-blog and contain traffic information fusion.