CN113139125A - User demand driven service matching method - Google Patents
User demand driven service matching method Download PDFInfo
- Publication number
- CN113139125A CN113139125A CN202110429741.1A CN202110429741A CN113139125A CN 113139125 A CN113139125 A CN 113139125A CN 202110429741 A CN202110429741 A CN 202110429741A CN 113139125 A CN113139125 A CN 113139125A
- Authority
- CN
- China
- Prior art keywords
- service
- user
- requirements
- matching
- users
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000008569 process Effects 0.000 claims description 17
- 238000011156 evaluation Methods 0.000 claims description 16
- 230000004044 response Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 2
- 239000000126 substance Substances 0.000 claims description 2
- 238000012163 sequencing technique Methods 0.000 abstract description 5
- 230000001502 supplementing effect Effects 0.000 abstract description 2
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000005284 excitation Effects 0.000 description 4
- 230000003247 decreasing effect Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- OOYGSFOGFJDDHP-KMCOLRRFSA-N kanamycin A sulfate Chemical group OS(O)(=O)=O.O[C@@H]1[C@@H](O)[C@H](O)[C@@H](CN)O[C@@H]1O[C@H]1[C@H](O)[C@@H](O[C@@H]2[C@@H]([C@@H](N)[C@H](O)[C@@H](CO)O2)O)[C@H](N)C[C@@H]1N OOYGSFOGFJDDHP-KMCOLRRFSA-N 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a service matching method, wherein the service adopts a service basic model, and the method comprises the following steps: attribute information of service name, type, field, label, operation, service quality, the method includes: receiving requirements input by a user, wherein the requirements at least comprise keywords, types, fields and labels of service names; matching similar users of the users, supplementing the labels of the users from the labels of the similar users, and giving weights to the supplemented labels to obtain a label weight set; and matching the services with the three types of requirements layer by layer, acquiring the matched services and sequencing the matched services according to the weight. In the input stage, the invention leads the user to input the requirements under guidance to replace the traditional service matching based on keywords, and adopts searching similar users to expand and correct the requirements of the user, classifies the requirements according to the priority, and matches the services layer by layer according to the required priority, thereby leading the service matching to be more accurate and providing high-quality service for the user.
Description
Technical Field
The invention relates to the field of Internet application, in particular to a service matching method.
Background
The Web service exists by relying on the Internet, is a platform-independent Web application with low coupling and self-contained characteristics, and can provide one or more functional services for a user at one time. With the rapid progress and deepening of application of modern information communication technology in China, the modern information technology in China accelerates the trend of changing to service. Information technologies such as cloud computing, internet of things, mobile internet, big data and the like are combined with traditional services, the scale and variety of the services tend to diversify, and the achievements promote the research of Web services on one hand and greatly increase the number of the services with similar functions on the other hand. The user needs to obtain a service meeting the requirement from a large amount of service information, and the user requirement becomes the key point of service matching attention in the process.
Service matching is one way to implement screening for large-scale services. Quantitative service is obtained through service matching, and time consumption of a user in service selection is reduced. However, the needs of users have dynamic, complex and susceptible characteristics when matching services, and most users lack professional domain knowledge, so that description of individual needs is often uncertain and may be different from general to detailed. Such situations are likely to occur in "new users" with less knowledge of the service, and their descriptions will be at a distance from "old users" who have some knowledge of the service or knowledge of the relevant Web service system.
For example, in the transportation aspect, some old users may put forward specific demands such as "route navigation, high-speed payment, road condition report", and new users with unclear demands may put forward ambiguous demands such as "transportation". Therefore, how to improve the problems existing in service matching from the user requirements, reduce the number of services by using the requirements, and reduce the selection time of the user for the services is a problem to be solved at present.
The expansion according to the above problem can be divided into three problems:
the method has the advantages that the method can normalize the user demand information and improve the accuracy of describing the personal demand information by the user.
At present, the majority of methods adopted in Web service portal websites mainly include: based on a keyword matching method. The method is simple to operate, low in calculation complexity and convenient to use. The method has strong interactivity with the user, and the user can organize the description sentences by himself for demand expression, but the user can not clearly describe various personal demands, and finally the matching result is influenced, so that the feedback service can not reach the expectation of the user.
Secondly, the user usually has difficulty in expressing the requirements completely and accurately, and how to select the attributes of the connectable service and the user from various requirement information of the user to expand the matching range and fill up and correct the requirements of the user.
In service matching, the user's preference information is part of the user's requirements. Based on the crowd psychology of the user, the preference needs of the user are mostly derived from the recommendations of other users. The research aiming at the problems adopts a collaborative filtering method. The method can effectively discover the personal preference of the user to meet the diversified requirements of the user. However, the variety of the preference information of the user is large, and recommendation of all the preference information cannot be realized, so that the user needs to consider which attribute information to select to help better comb personal requirements, expand the matching range and correct and complement the requirement information.
And thirdly, how to carry out priority division on various kinds of demand information of the user and realize the problem of step-by-step matching between the demand and the service.
In service matching, the user needs are many, but not every need is the most important to the user. Therefore, there is a need to consider how to implement hierarchical matching of demands, and improve the satisfaction and loyalty of users while achieving the basic demand information of users.
Disclosure of Invention
The present invention is directed to the above problem, and according to a first aspect of the present invention, a method for matching a service is provided, where the service uses a service basic model, and the service basic model includes: attribute information of service name, type, field, label, operation, service quality, the method includes:
step 100: receiving requirements input by a user, wherein the requirements at least comprise keywords, types and fields of service names;
step 310: matching the service according to the keyword, type and field of the service name input by the user to obtain a candidate service S1 { S1, S2, S3.., sx };
step 320: and matching the candidate service S1 with the tags and operations input by the user to obtain a candidate service S2 { S1, S2, S3.
In an embodiment of the present invention, in step 100, the requirement input by the user further includes tag operation and quality of service, and the method further includes:
step 200: matching similar users of the users, filling up the labels of the users from the labels of the similar users, giving weights to the filled-up labels, and obtaining a label weight set Iuser。
In one embodiment of the present invention, wherein the step 200 comprises: and calculating N users ui with similarity greater than a preset first threshold value with the user through cosine similarity, wherein i is 1 … N, and N is a non-negative integer, wherein the cosine similarity is calculated as the product of the number of the common labels of the two users divided by the number of the labels of the two users modulo.
In an embodiment of the present invention, wherein the step 200 further comprises: and counting the labels which are lacked when the user compares with the user ui, sequencing the lacked labels according to the recommendation degree, giving a weight TCount to the label with the highest recommendation degree, and sequentially decreasing the weights of the other labels, wherein TCount is the number of the lacked labels, and the weight TCount +1 is the original label of the user.
In one embodiment of the present invention, step 300 further comprises:
step 330: matching the candidate service S2 with the service quality input by the user to obtain a candidate service S3 { S1, S2, S3.
Step 340: services of S2 and S3 are sorted by weight.
In one embodiment of the present invention, the input requirements adopt a user requirements model, the user requirements model comprises basic requirements, expected requirements and exciting requirements, wherein the basic requirements comprise: service name, type and domain; the desired requirements include: labeling and operating; the excitement needs include: price, evaluation and response time;
wherein the content of the first and second substances,
step 310 includes: if the service name of the service contains a keyword of the basic requirement, or the service type is equal to the type of the basic requirement, or the field of the service is equal to the field of the basic requirement, the matching is successful, and all the successfully matched services form a candidate service S1 in the first stage { S1, S2, S3.., sx }; and
step 320 includes: matching and weight appending the service in S1, if the name of the service tag includes the name of the desired demand tag, adding the weight of the service, plus IuserThe weight of the tag in (1); if the operation of the service includes the operation desired by the user, the service weight is increased by 1, and all such services are taken as the candidate service of the second stage S2 { (S1, S2, S3.., sy }.
In one embodiment of the present invention, step 330 comprises: if the price of the service is equal to the price of the excitement demand, or the evaluation of the service is equal to the service evaluation of the excitement demand, or the service response time of the service is equal to the response time of the excitement demand, the weight of the service is added by 1, and all the services are used as candidate services of the third stage, namely S3 { S1, S2, S3.
In an embodiment of the present invention, step 340 further comprises: before sorting, whether the candidate service has a difference in the process of adding the weight twice is calculated, and if not, the service is removed.
According to a second aspect of the present invention, there is provided a computer readable storage medium, in which one or more computer programs are stored, which when executed, are for implementing the method of service matching of the present invention.
According to a third aspect of the invention there is provided a computing system comprising: a storage device, and one or more processors; wherein the storage means is adapted to store one or more computer programs which, when executed by the processor, are adapted to carry out the method of service matching of the present invention.
In the input stage, the invention leads the user to input the requirements under guidance to replace the traditional service matching based on keywords, and adopts searching similar users to expand and correct the requirements of the user, classifies the requirements according to the priority, and matches the services layer by layer according to the required priority, thereby leading the service matching to be more accurate and providing high-quality service for the user.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort. In the drawings:
FIG. 1 shows a service infrastructure model diagram according to an embodiment of the invention;
FIG. 2 illustrates a demand model diagram according to an embodiment of the present invention;
FIG. 3 illustrates a demand classification diagram according to an embodiment of the present invention;
FIG. 4 illustrates a service matching overall schematic diagram according to an embodiment of the invention;
fig. 5 shows a development diagram of the principle of service matching according to an embodiment of the present invention.
Detailed Description
In view of the problems presented in the background art, the inventors have conducted research and have presented a user demand-driven service matching method. In summary, the method first establishes a service base model and a demand model, and forms service data according to the service that can be provided, wherein the service data is in a form provided by the service base model. The method adopts the user requirement as input instead of the traditional keyword-based method, and the user inputs the requirement by guidance when inputting the requirement, so that the input requirement can adopt the form provided by a requirement model. The requirements for user input can be further modified and expanded. And matching the modified and expanded requirements with service data according to the type and priority of the requirements to form a service list which is required by the user and is sorted according to the weight.
The service matching of the present invention is an important step of service discovery, and the objects of action are users and services, respectively. The user is a demand input party of the whole service matching process, the demand information is complicated, and the attribute information of the service is lack of more multidimensional understanding, so that the time of the user in the process of carrying out demand description and service selection is increased. Therefore, in order to reduce the time of the user, the invention proposes the idea of user demand classification. On the basis of the idea, in order to better position the user requirements, optimize the service matching process, improve the satisfaction degree of the user on the service matching result, and finally provide the service matching method driven by the user requirements.
The service base model and the demand model used by the present invention are described below.
Service base model
The formal description of the service is the basis of service matching and is also the key for constructing a user demand model, and can be found according to personal demand information. The user has more uncertain factors when describing the personal requirement information. Therefore, in order to realize accurate matching of the requirements and the services, the attribute information of the services needs to be captured so as to support matching between the requirements of the users and the service information. For this purpose, a service infrastructure model is defined according to attribute information of the service, as shown in fig. 1. Meanwhile, in order to improve the accuracy of the user in inputting, the input and output of the service are defined in a set manner:
definition 1. service base model. And constructing a service model according to the user requirements, and defining the service as a six-tuple. The formalized representation of the service base model is:
S＝<Sname,type,field,Tags,OPSet,Qos>
wherein S isnameRepresenting a service name; type represents a service type; field represents the domain to which the service belongs; tags is a set of label sets for the service S, OPSet is a set of operations for the service, and Qos is a set of quality of service attribute information.
Tags＝{<Tagid,Tagname>}
Wherein, in the label setMay contain multiple tags, tagsidIs a serial number of a Tag, TagnameIs the name of a tag.
OPSet＝{<OPname,Input,Output>}
Wherein a service comprises a plurality of operations, OPnameThe name of a service operation is represented, Input is an Input parameter set corresponding to the operation, and Output is an Output parameter set corresponding to the operation.
Input＝{<Inname,Intype>}
Wherein Input represents a service operation OPnameIncluding the input parameter name InnameAnd type Intype。
Output＝{<Outname,Outtype>}
Wherein Output is a service operation OPnameIncluding the output parameter name OutnameAnd type Outtype。
The introduction of the non-functional attribute Qos can facilitate the matching of users to services. The Qos attributes are various and can be mainly classified into three types:
by service provider, such as service price, service authentication, authorization, etc.;
service consumers use the service and feedback information is obtained, such as service evaluation;
the services are provided by third parties other than the provider and the consumer during the runtime phase, e.g., service response time, availability, execution time, etc.
Combining the above three parts defines Qos as, for example, the following triplet.
Qos＝<price,evaluation,responseTime>
Where price is the service price, service evaluation and service response time responseTime.
Definitions 2. input output parameter terminology set for input and output of services, assume that there are n services s in the current domain1,s2,s3,...,snAny service siComprising a plurality of input/output parameters, according to the fieldThe inputs and outputs of the services in the domain are summarized to form a term set strerm in the corresponding domain:
STerm＝(∪ini)∪(∪outi)
the input and output set can be organized according to the service field, and when a user performs matching, the user can extract input and output from the specified term set according to the field, so that the matching efficiency is improved, and the input normalization of the user is ensured.
Demand model
The user demand model is based on the service basic model, and based on the demand type, the actual user demand is associated with the service attribute to promote the normalization of the user demand description and promote the matching of the service, so that as shown in fig. 2, the user demand model includes three parts including a basic demand, an expected demand and an exciting demand.
The requirement classification is to find that the requirements are different in priority when the requirements of users are researched, and the requirements are gradually diversified. Based on the Kano model proposed by Noriaki Kano in 1984, and taking service background into full consideration, user requirements are roughly divided into three categories: basic requirement (Basic Quality), expectation requirement (Performance Quality), and Excitement requirement (excitation Quality), as shown in fig. 3.
Basic requirements are as follows: the requirement is a basic requirement of the user for service matching, the requirement is considered to be necessarily met by the user, and if the requirement is met, the satisfaction degree of the user cannot be improved. If not, user satisfaction may decrease. Thus, the basic requirement set mainly corresponds to some basic attribute information of the service.
The expected demand is as follows: this type of demand is a demand attribute that is somewhat appealing to users, and user satisfaction is proportionally related to the fulfillment of that demand. The functional attribute information corresponding to the service is mainly concentrated in the desired demand set.
Exciting requirements: the requirements are requirement attributes with large attraction to users and are beneficial to improving the satisfaction degree of the users, so that the requirements of the users are classified mainly aiming at some non-functional attribute information of the services in the excitation requirement set.
As shown in FIG. 2, the demand model is defined as follows:
defining 3, a user requirement model, defining the user requirement model as a triple according to the requirement type, and formalizing the user requirement model as
UR＝<BQ,PQ,EQ>
Wherein BQ represents the basic requirement of the user, PQ represents the expected requirement of the user, and EQ represents the exciting requirement of the user.
BQ＝<Sname,type,field>
The information corresponding to the service basic attribute in the BQ is the service name SnameType, and domain field.
PQ＝<Tags,OPSet>
The information in the PQ corresponding to the service basic attribute is the tag Tags and the operation OPSet.
Tags＝{<Tagname>}
Wherein Tag only includes Tag name Tagname.
OPSet＝<Input,Output>
Wherein, the OPSet includes an Input parameter set Input and an Output parameter set Output.
Input＝<Inname,Intype>
Wherein, Input includes an Input parameter name InnameAnd type Intype.
Output＝<Outname,Outtype>
Wherein Output includes the Output parameter name OutnameAnd type Outtype.
EQ＝<price,evaluation,responseTime>
The information corresponding to the service basic attribute in the EQ is price, evaluation and response time responseTime.
The service matching overall schematic diagram is shown in fig. 4, and the whole process involves four stages, respectively: requirement input, requirement information correction, completion and classification, service matching and output.
1. Need toInput stage
In this phase, the user needs to enter all the requirements, including the service name SnameKeyword, type, field, tag, operation OPSet, and quality of service (Qos). According to one embodiment of the present invention, the keyword, type and domain information are required to be filled in, and the other information is filling information. The user may direct input by the name of the request.
2. Stage for correcting, supplementing and classifying demand information
The method mainly corrects, supplements and classifies information obtained in the input stage, obtains similar users through matching of a collaborative filtering algorithm by using label information obtained in the input stage, corrects and supplements the label information, and then classifies the demand information based on a constructed user demand model. In the module, the similarity between users is obtained by using a cosine similarity calculation formula, then the user most similar to the current user A is found, and the preference label of the user is obtained, wherein the preference label is derived from the past use record of the user, and the specific number of the preference labels can be set based on the actual application environment. After matching between users is completed, the user with the highest similarity to the user A is B, the preference labels of the B exist labels which are not added in the initial requirement description of the user A, the predicted preference values of the user A to the labels can be calculated, then a new recommended label set is formed by combining the labels in the original requirement of the user A, finally, the weight is given to each label, wherein the weight of the original label is the highest, and the candidate labels are assigned according to the height of the predicted preference value. According to an embodiment of the present invention, the specific process is as follows.
Suppose that the relevant service is matched in the background of the travel traffic field, user a is the current active user, and the requirement label of user a is a1,d2,d3,d4Selecting old users with more services in the field of travel traffic as oldUser (u)1,u2,u3,u4Each old user contains different numbersThe interest labels k of the old users are derived from records of services used by the old users, the first 5 labels with the largest number of use times are taken as preference labels, and the use time of part of users is short, the preference labels are few, so that the number k of the preference labels of each old user ranges from [1,5 ]]. The interest tag of the old user is denoted by t, and the threshold for matching is min threshold. And finally obtaining a group of label sets for matching the requirements with the services through similarity matching and recommendation degrees of the user a for the labels.
(1) Matching process of similar users
Firstly, the tag set selected by the user a is a.tags, the a.tags are from the tags input in the current demand input, if no tag is input, the similarity user matching is not performed, and the tag set corresponding to the old user is olduser. Establishing an inverted list between the tag and the user, as shown in table 2:
TABLE 1 user-tag initial correspondence table
| User' s | Label (R) |
| a | d1,d2,d3,d4 |
| u1 | d1,d2,d4,t1 |
| u2 | d1,d4,t2 |
| u3 | t3,t4 |
| u4 | d2,d3,t4 |
TABLE 2 TAG-USER INVERTED TABLE
| Label (R) | User' s |
| d1 | a,u1,u2 |
| d2 | a,u1,u4 |
| d3 | a,u4 |
| d4 | a,u1,u2 |
| t1 | u1 |
| t2 | u2 |
| t3 | u3 |
| t4 | u3,u4 |
Calculating user a and any user uiThe cosine similarity of (a) is shown in formula (1):
where N (a) is the set of tags for user a, N (u)i) Is user uiSet of labels of (1), Sim (a, u)i) Representing user a and user uiA similarity value of (a). The numerator is the number of the labels shared by the two users, the denominator is the product of the number of the labels of the two users, and the smaller the included angle is, the higher the similarity is.
And thirdly, verifying whether the similarity of the compared users is greater than min _ threshold, and if so, determining that the user is a similar user.
According to one embodiment of the invention, in the process of matching similar users, there is an unexpected condition that matching of similar users fails, and if the condition occurs, the service matching process is directly entered by using the label selected by the current user.
(2) Tag recommendation degree process
Obtaining N users similar to a according to the matching process of the similar users, and counting the a and the similar users uiCompare the missing tags tjAnd judging the user corresponding to the lacking label according to the matching process of the similar users, and calculating the recommendation degree of the user a to the lacking label, as shown in a formula (2):
where P (a, N) contains the N users closest to user a, and Q (j) is the pairThe label j has a set of users with preferences,
is user a and user uiThe similarity of (a. ui) is equal to Sim (a. ui),
is user uiInterest in tag j.
Secondly, adding weights to the labels according to the recommendation degree to obtain a final label weight set:
Iuser＝{<tag1,weight1>,<tag2,weight2>,...,<tagn,weightn>}
wherein IuserTo complement the revised set of tags, tagnIs the label name, weightnIs the corresponding weight of the label.
The weight is calculated in the following way: for the labels lacking in the user a, firstly, the number TCount of the labels lacking in the user a is obtained, then the labels are sequentially sorted according to the recommendation degree, and added to the set Irecommend＝{<t1,pt1>,...,<tj,ptj>,...,<tk,ptk>Where t isjIs a label p that user a lackstjIs the recommendation corresponding to the tag. And adding weights to the tags according to the sorting result, wherein the weight of the tag with the highest recommendation degree is TCount (positive type), and the weights of the rest tags are sequentially decreased. And if the recommendation degrees of the two labels are the same, adding the same weight. For the original tags of the user a, the weight is the highest value of the weights in the missing tags plus 1. For example, if the total number of missing tags is 5, the weight of the tag with the highest recommendation degree is 5, and then the tag is decreased once, if tag comes from user a, the weight of the original tag is 6.
3. Service matching phase
The stage is used for matching the classified demands and comprises a service matching module. An LMIA algorithm is provided in the module, and the main idea of the algorithm is to match basic, expectation and excitation requirements of the user in sequence according to the results after the requirements are classified and the priority of the requirements. There is a weighted assignment of both the desired demand and the excitement demand in the match. After matching all the requirements, the difference of the two rounds of weights needs to be compared, and the services generating the difference are output in an order of increasing amplitude. The specific flow of the module is shown as the service matching stage in the lower half of fig. 5.
Firstly, according to a user demand model, input contents of a user in actual matching are defined.
Defining 4, inputting actual matching by a user, and filling the required information according to a certain rule in the actual matching by the user, thereby defining a six-element group
R<keyword,type,field,Iuser,OPSet,Qos>.
Wherein, keyword represents the keyword of the query; type represents type; i isuserRepresenting the label, because in the matching process of similar users, the label requirement Tags of the users are replaced by IuserThus, the user needs I for the taguserRepresents; field represents the domain; the OPSet represents the operation, including the input and output of the service, whose source is STerm; qos represents the range of quality of service attributes that a user desires.
And filling information by a user, classifying the required information, and matching and sequencing the services according to the required types.
The LMIA algorithm consists of five main parts:
(1) and (3) classifying the requirements: the demand information of the user is divided according to a user demand model, and meanwhile, the tag information initially filled by the user is replaced by the tag set obtained in the matching process of similar users, so that a basic demand Ubq, an expected demand Upq and an exciting demand Ueq are obtained.
(2) Basic requirement matching: this stage is through Ubq<keyword,type,filed>The name and type of the service are matched. Due to the service name (S)name) The function of the service can be reflected, so that the keyword is matched with the name of the service, and the type and the field of the service are matched. If the service name contains a base requirementIf the keyword or the service type of the service is equal to the type of the basic requirement, or the field of the service is equal to the field of the basic requirement, the matching is successful, and all the successfully matched services are taken as the first stage candidate service S1 ═ S { (S)1,s2,s3,...,sn}。
(3) The expected demand matches: s1 ═ S obtained at this stage1,s2,s3,...,sxH, first by tag set:
Iuser＝{<tag1,weight1>,<tag2,weight2>,...,<tagk,weightn>}
matching and weight addition are performed for the service in S1, and the weight sum of the service is calculated. And then, the input and output of the service are matched, and the input and output information of the user for the service is extracted from the STerm, so that the matching time for the service is simplified. If the name of the service tag includes the name of the desired demand tag, the weight of the service is increased, plus IuserThe weight of the tag in (1). If the service OPSet contains the desired demand OPSet, the service weight is increased by 1, and all such services are taken as the candidate service for the second phase S2 ═ S1,s2,s3,...,sy}。
(4) Excitation requirement matching: and screening and sorting the service quality attributes according to the obtained S2. And comparing the Qos attribute value of each service with the user target expected value of the Qos to finally obtain a group of services meeting the conditions. If the price of the service is equal to the price of the excitement demand, or the service evaluation is equal to the evaluation of the excitement demand, or the response time of the service is equal to the response time of the excitement demand, the weight of the service is added by 1, and all the services are taken as candidate services of the third stage, namely S3 ═ S1,s2,s3,...,sz}。
(5) Service sequencing: and calculating whether the candidate services have a difference value in the process of adding the weights twice, if so, returning, and if not, removing, and sorting the returned services according to the weights.
The pseudo-code of the LMIA algorithm is described as follows:
inputting a user requirement R; a constructed user demand model UR; service data SD
Output service list US
1.BEGIN
2.List<Ubq,Upq,Ueq>＝reClassification(UR)；
3.// user demand Classification
4.FOR ALL Service s∈SD DO
5.IF s.Sname Contains(Ubq.keyword)||
6.s.Stype＝＝(Ubq.type)||
7.s.Sfield＝＝(Ubq.field)THEN
8.INSERT S1 END IF
9.END FOR
10.FOR ALL Service s∈S1 DO
11.IF s.Tags.tagName Contains(Upq.Tags.Tagname)THEN
12.s.weight+＝Upq.Iuser.weight END IF
13.// traverse OPSet decision
14.IF s.OPSet Contains(Upq.OPSet)THEN
Weight + 1// with both input and output increased by 1
16.INSERT S2 END IF
17.END FOR
18.FOR ALL Service s∈S2 DO
19.IF s.price＝＝Ueq.price||
20.s.evaluation＝＝Ueq.evaluation||
21.s.responseTime＝＝
22.Ueq.responseTime THEN
Weight + ═ 1// cost compliance 23.s.
24. The range of evaluation and corresponding time are all increased by 1
25.INSERT S3 END IF
26.END FOR
27.// S2 and S3 two weight comparisons
28.// drop service return US with no change
29.FOR ALL Service s2,s3∈S2,S3 DO
30.IF s3.weight≠s2.weight INSERT US
31.END IF END FOR
32.// sorting services in the US back by weight
33.Sort(US.weight,US)
34.RETURN US
35.END
4. Output stage
At this stage, the name S of the service is output in the form of a listnameService type, field to which the service belongs, tag name of the service tag. The user may make a selection of services for the output service.
The prior art methods for service matching mainly focus on semantic matching and improvement of service matching algorithms. The invention provides a service matching method based on user requirements by combining the characteristics of services. Firstly, similar users are found in a collaborative filtering mode, and the query tag set is expanded. Secondly, classifying the user requirements through a user requirement model, matching the services by utilizing an LMIA algorithm, adding weights, finally sequencing the services according to the weights, and outputting the services satisfied by the user.
The previous description is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to the disclosure will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Moreover, all or a portion of any aspect and/or embodiment may be utilized with all or a portion of any other aspect and/or embodiment, unless stated otherwise. Thus, the disclosure is not intended to be limited to the examples and designs described herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of service matching, the service employing a service base model, the service base model comprising: attribute information of service name, type, field, label, operation, service quality, the method includes:
step 100: receiving requirements input by a user, wherein the requirements at least comprise keywords, types and fields of service names;
step 310: matching the service according to the keyword, type and field of the service name input by the user to obtain a candidate service S1 ═ { S }1,s2,s3,...,sx}；
Step 320: matching the candidate service S1 with the label and operation input by the user to obtain a candidate service S2 ═ { S }1,s2,s3,...,sy}。
2. The method of claim 1, wherein the user input requirements further include label manipulation, quality of service, and the method further comprises:
step 200: matching similar users of the users, filling up the labels of the users from the labels of the similar users, giving weights to the filled-up labels, and obtaining a label weight set Iuser。
3. The method of claim 2, wherein the step 200 comprises: calculating N users u with similarity greater than a preset first threshold value to the user through cosine similarityiAnd i is 1 … N, wherein N is a non-negative integer, and the cosine similarity is calculated as the product of the number of tags shared by two users divided by the number of tags of the two users modulo.
4. The method of claim 3, wherein the step 200 further comprises: counting the users and the users uiComparing the lacking labels, sorting the lacking labels according to recommendation degree, giving weight TCount to the label with the highest recommendation degree, and sequentially giving weights of the other labelsAnd decrementing, wherein TCount is the number of missing tags and the user's original tags are weighted by TCount + 1.
5. The method of claim 2, step 300 further comprising:
step 330: matching based on the candidate service S2 and the service quality input by the user, and obtaining a candidate service S3 ═ { S }1,s2,s3,...,sz}；
Step 340: services of S2 and S3 are sorted by weight.
6. The method of claim 2, the input requirements employing a user requirements model, the user requirements model comprising base requirements, desired requirements, and excitement requirements, wherein base requirements comprise: service name, type and domain; the desired requirements include: labeling and operating; the excitement needs include: price, evaluation and response time;
wherein the content of the first and second substances,
step 310 includes: if the service name of the service contains the key word of the basic requirement, or the service type is equal to the type of the basic requirement, or the field of the service is equal to the field of the basic requirement, the matching is successful, and all the successfully matched services form a candidate service S1 ═ S { in the first stage1,s2,s3,...,sx}; and
step 320 includes: matching and weight appending the service in S1, if the name of the service tag includes the name of the desired demand tag, adding the weight of the service, plus IuserThe weight of the tag in (1); if the operation of the service includes the operation desired by the user, the service weight is increased by 1, and all the services are taken as the candidate service of the second stage S2 ═ S1,s2,s3,...,sy}。
7. The method of claim 5, step 330 comprising: if the price of the service is equal to the price of the excitement demand or the evaluation of the service is equal to the evaluation of the excitement demandOr the service response time of the service is equal to the response time of the excitement demand, the weight of the service is added with 1, and all the services are taken as the candidate services of the third stage, namely, S3 { S ═ S }1,s2,s3,...,sz}。
8. The method of claim 7, wherein step 340 further comprises: before sorting, whether the candidate service has a difference in the process of adding the weight twice is calculated, and if not, the service is removed.
9. A computer-readable storage medium, in which one or more computer programs are stored, which when executed, are for implementing the method of any one of claims 1-8.
10. A computing system, comprising:
a storage device, and one or more processors;
wherein the storage means is for storing one or more computer programs which, when executed by the processor, are for implementing the method of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110429741.1A CN113139125B (en) | 2021-04-21 | 2021-04-21 | User demand driven service matching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110429741.1A CN113139125B (en) | 2021-04-21 | 2021-04-21 | User demand driven service matching method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113139125A true CN113139125A (en) | 2021-07-20 |
| CN113139125B CN113139125B (en) | 2024-02-09 |
Family
ID=76813393
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110429741.1A Active CN113139125B (en) | 2021-04-21 | 2021-04-21 | User demand driven service matching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113139125B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114679498A (en) * | 2022-03-22 | 2022-06-28 | 亚信科技（中国）有限公司 | Intention information processing method and device, electronic equipment and readable storage medium |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103678431A (en) * | 2013-03-26 | 2014-03-26 | 南京邮电大学 | Recommendation method based on standard labels and item grades |
| CN104346408A (en) * | 2013-08-08 | 2015-02-11 | 中国移动通信集团公司 | Method and equipment for labeling network user |
| CN104915861A (en) * | 2015-06-15 | 2015-09-16 | 浙江经贸职业技术学院 | An electronic commerce recommendation method for a user group model constructed based on scores and labels |
| CN106484764A (en) * | 2016-08-30 | 2017-03-08 | 江苏名通信息科技有限公司 | User's similarity calculating method based on crowd portrayal technology |
| CN106776897A (en) * | 2016-11-29 | 2017-05-31 | 中国农业银行股份有限公司 | A kind of user's portrait label determines method and device |
| CN106909686A (en) * | 2017-03-06 | 2017-06-30 | 吉林省盛创科技有限公司 | A kind of man-machine interaction builds user's portrait cluster calculation method |
| CN108804564A (en) * | 2018-05-22 | 2018-11-13 | 深圳壹账通智能科技有限公司 | The combined recommendation method and terminal device of financial product |
| CN108985326A (en) * | 2018-06-05 | 2018-12-11 | 南京大学 | A kind of electric system demand response user's mark method based on cluster |
| CN109816233A (en) * | 2019-01-16 | 2019-05-28 | 云南电网有限责任公司信息中心 | User tag model generating method and device for power grid |
| CN110020094A (en) * | 2017-07-14 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of methods of exhibiting and relevant apparatus of search result |
| CN112001748A (en) * | 2020-08-14 | 2020-11-27 | 广州掌淘网络科技有限公司 | Data expansion method and equipment based on label propagation |
| CN112035750A (en) * | 2020-09-17 | 2020-12-04 | 上海二三四五网络科技有限公司 | Control method and device for user tag expansion |
| CN112364252A (en) * | 2021-01-12 | 2021-02-12 | 北京世纪好未来教育科技有限公司 | Content recommendation method and device, electronic equipment and storage medium |
| CN112598471A (en) * | 2020-12-25 | 2021-04-02 | 北京知因智慧科技有限公司 | Product recommendation method and device and electronic equipment |
-
2021
- 2021-04-21 CN CN202110429741.1A patent/CN113139125B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103678431A (en) * | 2013-03-26 | 2014-03-26 | 南京邮电大学 | Recommendation method based on standard labels and item grades |
| CN104346408A (en) * | 2013-08-08 | 2015-02-11 | 中国移动通信集团公司 | Method and equipment for labeling network user |
| CN104915861A (en) * | 2015-06-15 | 2015-09-16 | 浙江经贸职业技术学院 | An electronic commerce recommendation method for a user group model constructed based on scores and labels |
| CN106484764A (en) * | 2016-08-30 | 2017-03-08 | 江苏名通信息科技有限公司 | User's similarity calculating method based on crowd portrayal technology |
| CN106776897A (en) * | 2016-11-29 | 2017-05-31 | 中国农业银行股份有限公司 | A kind of user's portrait label determines method and device |
| CN106909686A (en) * | 2017-03-06 | 2017-06-30 | 吉林省盛创科技有限公司 | A kind of man-machine interaction builds user's portrait cluster calculation method |
| CN110020094A (en) * | 2017-07-14 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of methods of exhibiting and relevant apparatus of search result |
| CN108804564A (en) * | 2018-05-22 | 2018-11-13 | 深圳壹账通智能科技有限公司 | The combined recommendation method and terminal device of financial product |
| CN108985326A (en) * | 2018-06-05 | 2018-12-11 | 南京大学 | A kind of electric system demand response user's mark method based on cluster |
| CN109816233A (en) * | 2019-01-16 | 2019-05-28 | 云南电网有限责任公司信息中心 | User tag model generating method and device for power grid |
| CN112001748A (en) * | 2020-08-14 | 2020-11-27 | 广州掌淘网络科技有限公司 | Data expansion method and equipment based on label propagation |
| CN112035750A (en) * | 2020-09-17 | 2020-12-04 | 上海二三四五网络科技有限公司 | Control method and device for user tag expansion |
| CN112598471A (en) * | 2020-12-25 | 2021-04-02 | 北京知因智慧科技有限公司 | Product recommendation method and device and electronic equipment |
| CN112364252A (en) * | 2021-01-12 | 2021-02-12 | 北京世纪好未来教育科技有限公司 | Content recommendation method and device, electronic equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| HUIYING ZHANG等: "Research on User Demand-Driven Service Matching Methodology", 《2021 IEEE 6TH INTERNATIONAL CONFERENCE ON CLOUD COMPUTING AND BIG DATA ANALYTICS》, pages 1 - 6 * |
| XING XU等: "Cross-modal Retrieval with Label Completion", 《MM \'16》, pages 302 - 306 * |
| 查琇山: "基于缺失值补全和SVD的手游推荐方法", 《计算机应用研究》, vol. 2020, no. 02, pages 166 - 169 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114679498A (en) * | 2022-03-22 | 2022-06-28 | 亚信科技（中国）有限公司 | Intention information processing method and device, electronic equipment and readable storage medium |
| CN114679498B (en) * | 2022-03-22 | 2024-04-26 | 亚信科技（中国）有限公司 | Intention information processing method, device, electronic equipment and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113139125B (en) | 2024-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20230342634A1 (en) | System and/or method for generating clean records from imperfect data using model stack(s) including classification model(s) and confidence model(s) | |
| Gal | Uncertain schema matching | |
| US9613164B2 (en) | System and method for data management in large data networks | |
| US7962891B2 (en) | Automated method for structured artifact matching | |
| US20020120619A1 (en) | Automated categorization, placement, search and retrieval of user-contributed items | |
| US20090327249A1 (en) | Intellegent Data Search Engine | |
| CN110609902A (en) | Text processing method and device based on fusion knowledge graph | |
| Carmel et al. | Multi-objective ranking optimization for product search using stochastic label aggregation | |
| CN104239340A (en) | Search result screening method and search result screening device | |
| CN114218400A (en) | Semantic-based data lake query system and method | |
| CN106528648A (en) | Distributed keyword approximate search method for RDF in combination with Redis memory database | |
| CN102959539B (en) | Item recommendation method during a kind of repeat in work and system | |
| CN104778284A (en) | Spatial image inquiring method and system | |
| CN113139125A (en) | User demand driven service matching method | |
| CN104123285A (en) | Navigation method and device for search results | |
| Albertoni et al. | Asymmetric and context-dependent semantic similarity among ontology instances | |
| CN110717095B (en) | Service item pushing method and device | |
| CN116308683A (en) | Knowledge-graph-based clothing brand positioning recommendation method, equipment and storage medium | |
| Valliyammai et al. | An intelligent personalized recommendation for travel group planning based on reviews | |
| CN111651456B (en) | Potential user determination method, service pushing method and device | |
| CN109408713B (en) | Software demand retrieval system based on user feedback information | |
| CN111291904A (en) | Preference prediction method and device and computer equipment | |
| Diao et al. | Scientific workflow recommendation based on service knowledge graph | |
| WO2020008180A1 (en) | Method and system for integrating data sets | |
| Lacom et al. | A Research Framework for B2B Green Marketing Innovation: the Design of Sustainable Websites |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |