E N D
2. Web Service Search
13. How to Improve Web Service Search?
14. 1) Provide Similar WS Operations
15. 2) Provide Operations with Similar Inputs/Outputs
16. 3) Provide Composable WS Operations
19. Elementary Problems
20. Can We Apply Previous Work?
21. Why Text Matching Does not Apply?
23. Why Text Matching Does not Apply?
24. Operations Have More Complex Structures
25. Our Solution Part 1: Exploit Structure
26. Why Text Matching Does not Apply?
27. Parameter Names Are Highly Varied
28. Our Solution Part 2: Cluster Parameters into Concepts
29. Outline
30. Clustering Parameter Names
31. Criteria for an Ideal Clustering
32. Clustering Algorithm (I)
33. Clustering Algorithm (II)
34. Clustering Algorithm (III)
35. Clustering Algorithm (IV)
36. Clustering Algorithm (V)
37. Outlines
38. Experiment Data and Clustering Results
39. Example Clusters Mainly good
Tolerant of misspelling
Some noise
Merged after the fourth run
How many ops use itMainly good
Tolerant of misspelling
Some noise
Merged after the fourth run
How many ops use it
40. Example Clusters Mainly good
Tolerant of misspelling
Some noise
Merged after the fourth run
How many ops use itMainly good
Tolerant of misspelling
Some noise
Merged after the fourth run
How many ops use it
41. Example Clusters Mainly good
Tolerant of misspelling
Some noise
Merged after the fourth run
How many ops use itMainly good
Tolerant of misspelling
Some noise
Merged after the fourth run
How many ops use it
42. Measuring Top-K Precision
43. Top-k Precision for Operation Matching
44. Top-k Precision for Input/output Matching
45. Measuring Precision and Recall
46. Impact of Multiple Sources of Evidences in Operation Matching
47. Impact of Parameter Clustering in Input/output Matching
48. Conclusions