Skip to content

Latest commit

 

History

History
14 lines (12 loc) · 697 Bytes

README.md

File metadata and controls

14 lines (12 loc) · 697 Bytes

Ranking SVM

Rank each item by "pair-wise" approach

Implementation

  • item x: ("x.csv") x has feature values and a grade-level y (at the same row in "y.csv")
  • grade-level y: ("y.csv") y consists of grade (the first) and query id (the second)
  • one x or one y is one row in "csv" file
  • ranking SVM is implemented based on "pair-wise" approach
  • items are compared if items are in the same query id
  • this is implemented by using machine learning tool "scikit-learn"
  • (optional) pca for reducing feature dimension
  • (optional) baseline method for this task: finding most simimlar items (in test data) with items of high grade (in the train data)
  • train/test data: please refer to "main.py"