CN105930406A - Poisson decomposition based service recommendation method - Google Patents
Poisson decomposition based service recommendation method Download PDFInfo
- Publication number
- CN105930406A CN105930406A CN201610237950.5A CN201610237950A CN105930406A CN 105930406 A CN105930406 A CN 105930406A CN 201610237950 A CN201610237950 A CN 201610237950A CN 105930406 A CN105930406 A CN 105930406A
- Authority
- CN
- China
- Prior art keywords
- service
- services composition
- theme feature
- theme
- recommendation method
- 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
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
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a Poisson decomposition based service recommendation method. Three pieces of topic distribution on a Web service are obtained respectively by utilizing a descriptive text of the Web service, a historical call record of the Web service and evaluation of a user to the Web service, and three topic distribution results are fused as topic distribution of the Web service; and a time sequence of service combination is generated by utilizing release time information of existing service combination. When a developer proposes a development demand, a demand text proposed by the developer is analyzed to obtain topic distribution of new service combination; the topic distribution of the new service combination, the topic distribution of the service and the time sequence of the service combination are synthesized; and combined probability distribution of "service combination-service" is calculated. A Web service list demanded by the developer is obtained, an order of recommendation from high to low is represented with an order of probability values from big to small, and finally the recommended Web service list is provided for the user.
Description
Technical field
The present invention is the service recommendation method of a kind of service-oriented combination and exploitation person.The method utilizes the history of service
Use record, describe information and the user review information to service, use Poisson to decompose the theme of the service of extraction
Feature, excavates the matching grating of service and Services Composition, final proposition Poisson service recommendation method from theme aspect.
This method belongs to computer system modeling and data analysis field.
Background technology
Along with Service-Oriented Architecture Based (Service Oriented Architecture, hereinafter referred to as SOA) is wide
General application, internet is experiencing by the transformation of " data grid technology " to " service-centric ".On internet
A large amount of software suppliers change the rotating cylinder management mode of oneself, i.e. service (Software as a with software
Service, be called for short SaaS) pattern, the product of oneself is deployed on internet with the form of Web service.
Meanwhile, developer utilize on internet the open service of a large amount of SaaS patterns to develop the application of oneself,
Suitable Web service is embedded the program of oneself to reach quickly to develop and the purpose of convenient use.In this mistake
Cheng Zhong, numerous Web services the individual form person of being developed with dynamic combined use, and defines Services Composition
Or mashup, thus produce increase in value.
The transformation of new software development model also brings new problem, and on the one hand software supplier is on the internet
Having provided the user all kinds of Web services of magnanimity, this serve individual Various Functions, service quality is the most not to the utmost
It is identical, even similar Web service also certainly exists trickle difference between them.This makes
Developer selects suitable Web service process on stream and becomes very long and loaded down with trivial details, over time send out
The quantity of exhibition Web service is also quickly increasing so that this process is more complicated.On the other hand, developer
The functional requirement of exploitation software application is often complicated and changeable, and user is difficult to which speciality statement clearly needs
Web service.
With this problem of actual example tool.Such as, developer to develop such a application: " base
In the social networks application of GPS location, user can share geography information with the good friend in social networks ".?
During this section describes, developer's demand in fact adheres to the Web service support in three fields separately.One is that " map is with fixed
Position " service that field is relevant, in order to obtain geography information;Two is the relevant clothes in " social network-i i-platform " field
Business, in order to provide the interface connecting social network-i i-platform;Three is the related service in " mobile terminal " field, because of
' GPS location ' for user is likely to call the service in the relevant field of some mobile terminals.Then develop
In person needs to find these three field, suitably Web service is individual, uses in the development plan of oneself.?
During this, the technical problem that developer faces has two: one to be can to refine accurately and summarize out
Send out the field (generation referred to as theme in method) planning required Web service;Another one is can be in this sea, field
The Web service of amount is found the serve individual of suitable and definite needs.The two process obviously has certain
Difficulty, and be greatly increased the construction cycle and increase development cost.Therefore, transship at this information on services, again
In the case of lacking unified information Description standard, how to utilize semantic description information pointer that Services Composition is carried out effectively
Service recommendation, make user carry out Services Composition efficiently, the benign development to internet has highly important
Meaning.
Summary of the invention
In order to solve above-mentioned technical problem targetedly, the present invention proposes a kind of service decomposed based on Poisson and pushes away
Recommend method.When Services Composition developer proposes the functional requirement of new Services Composition, utilize this method extraction service
Theme feature, provides related service list for developer, the effective development time shortening Services Composition, reduces
Construction cycle.The method has taken into account speed and the accuracy requirement of service recommendation, achieves on True Data collection
Preferably effect.
Present invention firstly provides a kind of service recommendation method decomposed based on Poisson.The whole calculation of the inventive method
Method flow process, is made up of two parts: 1) model generation process;2) service recommendation process
1. model generation process
Model generation process comprises three subs:
A) the theme feature extraction serviced
I. from the description text extraction service theme feature of service.Obtain the Technique Using Both Text information of service, and profit
(Poisson Factorization, PF) is decomposed by the Technique Using Both Text information MAP of each service to by Poisson
On the vector of regular length, with matrix method, the theme feature distribution of service is carried out formalized description.
Ii. the history from service calls record extraction service theme feature.From calling of service, record extracts clothes
The theme feature distribution of business.
Iii. from the evaluation of user, extract service theme feature.PF algorithm is used to extract from service evaluation
The theme feature distribution of service.
B) theme feature serviced merges
Generating three theme feature distributions about service in the upper stage respectively, this stage uses CMF
Described three theme features distribution is merged by (Collective matrix factorization) algorithm.
C) time series of existing Services Composition generates
According to the issuing time order of Services Composition, according to the method for time slice, Services Composition is grouped, generates
The time series of Services Composition.
2. service recommendation process
Service recommendation process comprises two subs:
A) the theme feature extraction of Services Composition
Acquisition demand text, uses PF algorithm to extract the theme feature of destination service combination.
B) service list is recommended
The theme feature of integrated service combination, the theme feature of service and the time serial message of Services Composition,
Using the size order of probability distribution value as order standard, finally return that service recommendation list.
Summary content, method proposed by the invention and existing web service recommendation method (as
Mashup-description-based Collaborative Filtering(MDCF)、Time-aware
Collaborative Domain Regression (TCDR) etc.) to compare, the present invention service of fully excavating uses
The multiple data messages such as record, service text description, user's evaluation, it is ensured that the accuracy of recommendation;Next makes
By the PF mathematical method being more suitable for long-tail data, it is ensured that the real-time of recommendation, before there is good application
Scape.
Accompanying drawing explanation
Service recommendation method is had been described in detail by the application by accompanying drawing, and these descriptions are merely to illustrate this
Bright content, is not intended to limit the present invention.
Fig. 1 is the service recommendation method flow process decomposed based on Poisson in the present invention.
Fig. 2 is model generation process diagram in the present invention.
Fig. 3 is service recommendation process diagram in the present invention.
Detailed description of the invention
Below in conjunction with accompanying drawing, the present invention made the most concrete description.The table 1 of the end of writing relates to the present invention
The concrete meaning of middle symbol, and table 2 is for describing the parameter equation that the present invention relates to.
Fig. 1 describes the service recommendation method flow process decomposed based on Poisson.The method flow process includes extracting Web
The history of service calls record, the text of service describes and the user comment information of service, comprehensively generates service
Theme distribution;Issuing time section according to existing Services Composition generates the time series of Services Composition.In exploitation
When person proposes exploitation demand, method generates the theme distribution of Services Composition, then closes with the result of model generation process
The service recommendation list that applicable user requires is given after one-tenth.
The method of the present invention is made up of two stages in order, and first stage is model generation, generates Web
The theme distribution of service and the time series of Services Composition etc. are to be called;Second stage is service recommendation,
When having developer to propose exploitation demand, demand is combined with the model calculation, provide service recommendation sequence knot
Really.
Fig. 2 describes model generation process.Decompose including application Poisson, call note from the history of Web service
The user comment information that record, the text of service describe and service extracts the theme distribution of service, utilizes collaborative
Matrix disassembling method, adds time serial message, finally gives the theme distribution data of Web service.
Wherein model generation process comprises three subs:
A) the theme feature extraction serviced.From the description text extraction service theme feature of service.Each Web service
The explanation document that service function can be described by developer when issuing.This stage obtains the synthetic language of service
Justice information, and utilize Poisson to decompose (Poisson Factorization, PF) by the Technique Using Both Text letter of each service
Breath is mapped on the vector of a regular length, formalizes the theme feature distribution of service with matrix method
Describe.Record extraction service theme feature is called from the history of service.The outstanding Web service of part by
Existing Services Composition (mahsup) calls, and algorithm extracts the theme feature of this type of service from calling record
Distribution.Service theme feature is extracted from the evaluation of user.User often sends out after browsing, using Web service
Table some for the experience information of service or suggestion, the same master using PF algorithm to extract service from evaluate
Topic feature distribution.
B) theme feature serviced merges. and the upper stage generates three theme features about Web service respectively and divides
Cloth, three features distributions are merged by this stage use CMF (Collective matrix factorization) algorithm.
C) time series of existing Services Composition generates.According to the issuing time order of Services Composition, according to the time
Services Composition is grouped by the method for segmentation, generates the time series of Services Composition.
Fig. 3 describes service recommendation process.Decompose including application Poisson, the demand literary composition proposed from developer
Extract the theme distribution of Services Composition (mashup) in Ben, comprehensively by during model generation to Web
The theme distribution data of service, finally provide the recommendation list of Web service.
Service recommendation process comprises two subs:
A) the theme feature extraction of Services Composition.Use the demand text that developer proposes, use PF algorithm to carry
Take out the theme feature of destination service combination.
B) service list is recommended.The theme feature of integrated service combination, the theme feature of service and service group
The time serial message closed, using the size order of probability distribution value as order standard, finally returns that Web takes
Business recommendation list.
Concrete enforcement step is as follows
Step 0: come into effect;
Step 1~10 completes model generation process
Step 1: empirically determined βw,k、μs,k、δs,k、∈s,kAnd ηm,kThe yardstick of five Gamma distributions
Parameter (rte) and form parameter (shp).This step be arrange the theme feature of word, the description text of service,
The theme Gamma distribution of the service after record and fusion is called in user's evaluation of service, the history of service
Basic parameter, determines the original shape that Gamma is distributed.
Step 2: initialize the initial value of Gamma distribution with random value.The effect of this step is to arrange at the beginning of iteration
Value.
Step 3: according to table 2, Web service is described to the counting v making wordsw, work as vswDuring > 0, make
Use iterative formula undated parameter.This step effect is the Poisson distribution updating service describing word
Step 4: according to table 2, makes the counting c of word for the user comment of Web servicesw, work as csw> 0
Time, use iterative formula undated parameter.This step effect is the Poisson distribution updating user comment word
Step 5: according to table 2, text is described for Services Composition and makes the counting w of wordmw, work as wmw> 0
Time, use iterative formula undated parameter.This step effect is that the description text word Poisson updating Services Composition is divided
Cloth
Step 6: according to table 2, the history that Services Composition calls service calls record rms, work as rms> 0 makes
Use iterative formula undated parameter.This step effect is to update the Poisson distribution that service history is called
Step 7: according to table 2, use the v after updatingsw、csw、wmwAnd rms, according to iterative formula again
Calculate Gamma distribution parameter, and.This step is the theme the iteration of feature.
Step 8: according to the result of calculation of step 7, recalculate, i.e. recalculate the master of the service after fusion
Topic Gamma distribution.This step is that the theme feature of service merges iteration.
Step 9: repeat step 1~8 until restraining.So far the theme feature completing to service extracts and services
Theme feature merges.
Step 10: calculate time series, when being monthly divided into some according to the time distribution order of Services Composition
Between section, for time period t, the issue month of Services Composition m is tm, corresponding time sequential value is Tm,
FormulaWherein λη、λtFor coefficient, ληCan value 1, λtDesirable
Value 0.08, can suitably adjust.tcurrentFor current time.The time series that this step i.e. has Services Composition is raw
Become.
Step 11~15 completes service recommendation process.
Step 11: the scale parameter (rte) being distributed with the Services Composition Gamma of model generation process and shape
Parameter (shp) is as the parameter of the Gamma distribution initializing the requirement documents that developer proposes, with random value
Initialize the Gamma distribution of new Services Composition.
Step 12: utilize model generation process result of calculation βw,k, work as wmwDuring > 0, use repeatedly according to table 2
For formula undated parameter.This step is to calculate the Poisson distribution of developer's required Services Composition word.
Step 13: use formulaMore newly developed
The parameter of person requirement documents Gamma distribution.This step is the theme feature iteration of Services Composition.
Step 14: repeat 12~13, until convergence.So far the Services Composition feature extraction of user's request is completed.
Step 15: the new Services Composition m of developer meets Poisson distribution with service sUtilize formulaCalculate,It is normalized place
I.e. obtaining the joint probability distribution of " Services Composition Web service " after reason, this value is as final service recommendation
The results list is supplied to developer, and the biggest Web service representing correspondence of numerical value is more suitable for new Services Composition
m。
For the ease of statement, the symbol related in concrete steps definition is summarized as follows:
Table 1 symbol definition table
Table 2 parameter iteration formula
Claims (6)
1. the service recommendation method decomposed based on Poisson, it is characterised in that described recommendation method includes two processes:
First process is model generation process, and the time series of the theme distribution and Services Composition that generate service waits
Call;Second process is service recommendation process, when proposing demand, demand is tied mutually with the model calculation
Close, provide service recommendation ranking results;
Described first process comprises the following steps:
A) the theme feature extraction step serviced,
B) the theme feature fusion steps serviced,
C) the time series generation step of existing Services Composition;
Described second process comprises the following steps:
A) the theme feature extraction of Services Composition,
B) service list is recommended.
Service recommendation method the most according to claim 1, theme feature extraction step a) serviced includes:
I. from the description text extraction service theme feature of service,
Ii. the history from service calls record extraction service theme feature,
Iii. from the evaluation of user, extract service theme feature.
Service recommendation method the most according to claim 1 and 2, theme feature fusion steps b) serviced
Including: use CMF (Collective matrix factorization) algorithm by described three theme spies
Levy distribution to merge.
Service recommendation method the most according to any one of claim 1 to 3, c) has Services Composition
Time series generation step includes: according to the issuing time order of Services Composition, will according to the method for time slice
Services Composition is grouped, and generates the time series of Services Composition.
Service recommendation method the most according to any one of claim 1 to 4, wherein said a) Services Composition
Theme feature extraction include: obtain demand text, use PF algorithm extract destination service combination theme
Feature.
Service recommendation method the most according to any one of claim 1 to 5, described b) service list recommends step
Suddenly include: the theme feature of integrated service combination, the theme feature of service and the time series letter of Services Composition
Breath, using the size order of probability distribution value as order standard, finally returns that Web service recommendation list.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610237950.5A CN105930406B (en) | 2016-04-15 | 2016-04-15 | A kind of service recommendation method decomposed based on Poisson |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610237950.5A CN105930406B (en) | 2016-04-15 | 2016-04-15 | A kind of service recommendation method decomposed based on Poisson |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105930406A true CN105930406A (en) | 2016-09-07 |
CN105930406B CN105930406B (en) | 2019-03-22 |
Family
ID=56838272
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610237950.5A Active CN105930406B (en) | 2016-04-15 | 2016-04-15 | A kind of service recommendation method decomposed based on Poisson |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105930406B (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020107853A1 (en) * | 2000-07-26 | 2002-08-08 | Recommind Inc. | System and method for personalized search, information filtering, and for generating recommendations utilizing statistical latent class models |
CN102750360A (en) * | 2012-06-12 | 2012-10-24 | 清华大学 | Mining method of computer data for recommendation systems |
CN103473291A (en) * | 2013-09-02 | 2013-12-25 | 中国科学院软件研究所 | Personalized service recommendation system and method based on latent semantic probability models |
CN103544623A (en) * | 2013-11-06 | 2014-01-29 | 武汉大学 | Web service recommendation method based on user preference feature modeling |
CN103577414A (en) * | 2012-07-20 | 2014-02-12 | 富士通株式会社 | Data processing method and device |
CN103761254A (en) * | 2013-12-26 | 2014-04-30 | 清华大学 | Method for matching and recommending service themes in various fields |
US20150293978A1 (en) * | 2014-04-11 | 2015-10-15 | Baidu Online Network Technology (Beijing) Co., Ltd | Interactive searching and recommanding method and apparatus |
-
2016
- 2016-04-15 CN CN201610237950.5A patent/CN105930406B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020107853A1 (en) * | 2000-07-26 | 2002-08-08 | Recommind Inc. | System and method for personalized search, information filtering, and for generating recommendations utilizing statistical latent class models |
CN102750360A (en) * | 2012-06-12 | 2012-10-24 | 清华大学 | Mining method of computer data for recommendation systems |
CN103577414A (en) * | 2012-07-20 | 2014-02-12 | 富士通株式会社 | Data processing method and device |
CN103473291A (en) * | 2013-09-02 | 2013-12-25 | 中国科学院软件研究所 | Personalized service recommendation system and method based on latent semantic probability models |
CN103544623A (en) * | 2013-11-06 | 2014-01-29 | 武汉大学 | Web service recommendation method based on user preference feature modeling |
CN103761254A (en) * | 2013-12-26 | 2014-04-30 | 清华大学 | Method for matching and recommending service themes in various fields |
US20150293978A1 (en) * | 2014-04-11 | 2015-10-15 | Baidu Online Network Technology (Beijing) Co., Ltd | Interactive searching and recommanding method and apparatus |
Non-Patent Citations (1)
Title |
---|
HAO MA等: "Probabilistic Factor Models for Web Site Recommendation", 《SIGIR》 * |
Also Published As
Publication number | Publication date |
---|---|
CN105930406B (en) | 2019-03-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105224377B (en) | A kind of method and device by metadata automatic generating software project code file | |
Moon et al. | An approach to developing domain requirements as a core asset based on commonality and variability analysis in a product line | |
CN108984895B (en) | Model automatic identification method based on XML heterogeneous platform | |
CN110717320A (en) | Form/report designer and method suitable for multiple platforms and information management system | |
Josélyne et al. | Partitioning microservices: A domain engineering approach | |
CN107766042A (en) | A kind of visual data manager and device of the model file based on CSAR | |
CN111937027A (en) | Information processing apparatus, information processing method, and information processing program | |
Ewart et al. | Heritage building information modelling (HBIM): A review of published case studies | |
US11120200B1 (en) | Capturing unstructured information in application pages | |
CN108536718A (en) | A kind of method and system for the IT application in management realized based on input and output semantization | |
Beuche | Using pure: variants across the product line lifecycle | |
CN110032717B (en) | One-stop process business standardization design method | |
Kalenkova et al. | E-government services: Comparing real and expected user behavior | |
CN111160638A (en) | Conversion estimation method and device | |
Park et al. | Towards reliable business process simulation: a framework to integrate ERP systems | |
CN105894179A (en) | Business state transition method and system based on dynamic planning | |
CN105930483B (en) | Format Object generation method, apparatus and system | |
Acheson | Methodology for object-oriented system architecture development | |
CN105930406A (en) | Poisson decomposition based service recommendation method | |
Ahmad et al. | A FRAMEWORK FOR THE ADOPTION OF RAPID PROTOTYPING FOR SMEs: FROM STRATEGIC TO OPERATIONAL. | |
CN114124977A (en) | Cross-tenant data sharing method and device and electronic equipment | |
Oliveira et al. | Managing costs in software development | |
Susanty et al. | Design the geographical information system for supplier selection in batik industry | |
CN116204267B (en) | Method and device for generating intellectual property flow form | |
CN109711651A (en) | A kind of implicit factor model prestige visitor task recommendation method with modification vector |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |