Web Service Search

1
Web Service Search

• Xin Dong, Alon Halevy, Jayant Madhavan, Ema
  Nemes, Jun Zhang
• CSE Database Group
• University of Washington

2
Outline

• What are web services?
• How do we search web services now?
• How can we improve web service search?
• Why is the problem hard?
• How do we solve the problem?
• Woogle web service search engine

3
What is a Web Service?

• A web service is a software component, published
  and invoked across the Web, using standard
  XML-based protocols
• Web service functionality and interface are
  formally described in WSDL files
• Web services are published by registering WSDL
  URL and brief description in UDDI business
  registries.
• Offer large-scaled information sharing and
  ubiquitous computing

4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
How to Search Web Services?
8
Experience I
9
Experience II
10
Experience II
11
Experience II
12
Experience II
13
Experience II
14
Experience III
15
Experience III
16
Experience III
17
Experience III
18
Experience III
19
How to Improve Web Service Search?

• Operation is the unit for remote invocation
• Provide similar operations for a given operation

20
1) Provide Similar WS Operations

• Op1 GetTemperature
• Input Zip, Authorization
• Output Return
• Op2 WeatherFetcher
• Input PostCode
• Output TemperatureF, WindChill, Humidity

Similar Operations ? Selection
21
2) Provide Operations with Similar Inputs/Outputs

• Op1 GetTemperature
• Input Zip, Authorization
• Output Return
• Op2 WeatherFetcher
• Input PostCode
• Output TemperatureF, WindChill, Humidity
• Op3 LocalTimeByZipcode
• Input Zipcode
• Output LocalTimeByZipCodeResult
• Op4 ZipCodeToCityState
• Input ZipCode
• Output City, State

Similar Inputs ? Aggregation
22
3) Provide Composable WS Operations

• Op1 GetTemperature
• Input Zip, Authorization
• Output Return
• Op2 WeatherFetcher
• Input PostCode
• Output TemperatureF, WindChill, Humidity
• Op3 LocalTimeByZipcode
• Input Zipcode
• Output LocalTimeByZipCodeResult
• Op4 ZipCodeToCityState
• Input ZipCode
• Output City, State
• Op5 CityStateToZipCode
• Input City, State
• Output ZipCode

Input of Op2 is similar to Output of Op5 ?
Composition
23
Our Goal

• Given an operation, provide similar operations
• Operations with similar functionality (input and
  output)
• Operations with similar inputs/outputs
• Operations that can compose with the given
  operation
• Goal
• High recall Return potentially similar
  operations
• Good ranking Rank closer operations higher
• Operation matching problemInput/output matching
  problem

24
Why is the Problem Hard?

• Thousands of web servicesv.s. Billions of
  webpages? Return all potentially similar
  operations

25
Why is the Problem Hard?

• Thousands of web servicesv.s. Billions of
  webpages? Return all potentially similar
  operations
• Very brief description a couple of sentences or
  paragraphsv.s. Webpages are much longer? Lack
  of information

26
(No Transcript)
27
Why is the Problem Hard?

• Thousands of web servicesv.s. Billions of
  webpages? Return all potentially similar
  operations
• Very brief description a couple of sentences or
  paragraphsv.s. Webpages are much longer? Lack
  of information
• Operation and parameter names are highly varied?
  Finding word usage patterns are harder

28
Various Parameter Names

• Op1 GetTemperature
• Input Zip, Authorization
• Output Return
• Op2 WeatherFetcher
• Input PostCode
• Output TemperatureF, WindChill, Humidity
• Op3 LocalTimeByZipcode
• Input Zipcode
• Output LocalTimeByZipCodeResult
• Op4 ZipCodeToCityState
• Input ZipCode
• Output City, State
• Op5 CityStateToZipCode
• Input City, State
• Output ZipCode

29
Why is the Problem Hard?

• Thousands of web servicesv.s. Billions of
  webpages? Return all potentially similar
  operations
• Very brief description a couple of sentences or
  paragraphsv.s. Webpages are much longer? Lack
  of information
• Operation and parameter names are highly varied?
  Finding word usage patterns are harder
• Operations have more complex structurev.s. Web
  pages are mainly plain text? Finding term
  frequency is not enough

30
Different Structure

• Op1 GetTemperature
• Input Zip, Authorization
• Output Return
• Op2 WeatherFetcher
• Input PostCode
• Output TemperatureF, WindChill, Humidity
• Op3 LocalTimeByZipcode
• Input Zipcode
• Output LocalTimeByZipCodeResult
• Op4 ZipCodeToCityState
• Input ZipCode
• Output City, State
• Op5 CityStateToZipCode
• Input City, State
• Output ZipCode

Similar use of words, but opposite functionality
31
Naive Methods

• Match terms in operation functionality
  descriptions
• TF/IDF weights
• Considering information from other sources
  together as a bag of words
• Input/output parameters
• Web service descriptions

32
Our Strategy(1) Exploit Structure Information

• Compare each component separately
• host web services
• operation functionalities
• Inputs
• outputs
• Combine the results by setting different weights
  to different components

33
Our Strategy v.s. Naïve Methods
34
Our Strategy v.s. Naïve Methods
35
Our Strategy(2) Clustering Parameters into
Concepts

• Heuristics Parameters occurring together tend to
  express the same concepts
• Strategy Cluster parameters into concepts based
  on their co-occurrences
• Compare parameters and concepts separately, and
  combine the results.

36
Our Strategy v.s. Naïve Methods
37
Woogle

• A collection of 790 web services431 active web
  services, 1262 operations
• Version1.0
• Web service category browse
• Keyword search on web service descriptions
• Keyword search on inputs/outputs
• Web service on-site try
• Web service status report

38
(No Transcript)
39
(No Transcript)
40
(No Transcript)
41
(No Transcript)
42
(No Transcript)
43
Woogle

• Version 2.0
• Provide similar operations
• Web service template searchInput
  zipOutput city Functionality
  address transformReturn
  ZipCodeToCityState GetCity

44
Woogle

• Version 3.0
• Automatic web service compositionInput
  cityOutput
  temperatureFunctionality get weather
• ReturnCityStateToZipCode GetTemperatureCitySta
  teToZipCode GetWeather

45
Conclusions

• Propose new web service search interface
• Provide Lists of Similar Operations
• Strategies
• Exploit structure information
• Cluster parameters into concepts
• Woogle
• http//barb.cs.washington.edu8080/won/wonServlet

46
Conclusions

• Propose a new search method
• Search by parameters
• Define two fundamental problems
• Input-to-input matching
• Keyword-to-input matching
• Matching Algorithm Corpus-based matching
• Experimental results show that it has the
  potential to return a list of appropriate web
  services
• Woogle
• http//barb.cs.washington.edu8080/won/wonServlet