Uploaded image for project: 'Terrier Core'
  1. Terrier Core
  2. TR-223

Refactoring/Cleaning up of the package org.terrier.matching.models

          Details

          • Type: Improvement
          • Status: Resolved
          • Priority: Major
          • Resolution: Fixed
          • Affects Version/s: 3.5
          • Fix Version/s: 3.6
          • Component/s: .matching
          • Labels:
            None

            Description

            There are some inconsistencies in the package org.terrier.matching.models:
            - style guide rules (lines too long, empty lines full of tabs, unreadable score definitions)
            - final methods that prevent inheritance (BM25, TF_IDF)
            - missing library to standardize tf normalizations (sublinearity, length normalization...)
            - log definition in the Idf class that should be deprecated and moved to the new library

              Attachments

                Activity

                Ascending order - Click to sort in descending order
                Hide
                Permalink
                frousseau Francois Rousseau added a comment - - edited

                Patch related to the issue. Aiming at version 4.0

                ant test passes successfully.

                MAP on disks 4&5 (same as current version 3.5) for:

                • BB2 (0.2412) c=1.0
                • BM25 (0.2502) c=0.3444
                • DFI0 (0.2390)
                • DFR_BM25 (0.2043) c=0.3444
                • DFRee (0.2488)
                • DirichletLM (0.2324) mu=2500
                • DLH (0.2368)
                • DLH13 (0.2437)
                • DPH (0.2479)
                • Hiemstra_LM (0.2167) c=0.15
                • IFB2 (0.2400) c=1.0
                • In_expB2 (0.2408) c=1.0
                • In_expC2 (0.2336) c=1.0
                • InB2 (0.2392) c=1.0
                • InL2 (0.2405) c=1.0
                • Js_KLs (0.2504)
                • LemurTF_IDF (0.2273)
                • LGD (0.2484) c=1.0
                • PL2 (0.2538) c=9.15
                • TF-IDF (0.2508) c=0.3444
                • XSqrA_M (0.2477)
                Show
                frousseau Francois Rousseau added a comment - - edited Patch related to the issue. Aiming at version 4.0 ant test passes successfully. MAP on disks 4&5 (same as current version 3.5) for: BB2 (0.2412) c=1.0 BM25 (0.2502) c=0.3444 DFI0 (0.2390) DFR_BM25 (0.2043) c=0.3444 DFRee (0.2488) DirichletLM (0.2324) mu=2500 DLH (0.2368) DLH13 (0.2437) DPH (0.2479) Hiemstra_LM (0.2167) c=0.15 IFB2 (0.2400) c=1.0 In_expB2 (0.2408) c=1.0 In_expC2 (0.2336) c=1.0 InB2 (0.2392) c=1.0 InL2 (0.2405) c=1.0 Js_KLs (0.2504) LemurTF_IDF (0.2273) LGD (0.2484) c=1.0 PL2 (0.2538) c=9.15 TF-IDF (0.2508) c=0.3444 XSqrA_M (0.2477)
                Hide
                Permalink
                frousseau Francois Rousseau added a comment -

                New patch. For the tf normalizations's library, it's actually concave and not sublinear that we want.

                Show
                frousseau Francois Rousseau added a comment - New patch. For the tf normalizations's library, it's actually concave and not sublinear that we want.
                Hide
                Permalink
                richardm Richard McCreadie added a comment - - edited

                Integrated patch into the Terrier 3.5 build.

                Contains many javadoc additions and refactoring of IDF functions into a new WeightingModelLibrary class.

                Show
                richardm Richard McCreadie added a comment - - edited Integrated patch into the Terrier 3.5 build. Contains many javadoc additions and refactoring of IDF functions into a new WeightingModelLibrary class.
                Hide
                Permalink
                richardm Richard McCreadie added a comment -

                Committed version 3759.

                Show
                richardm Richard McCreadie added a comment - Committed version 3759.
                Hide
                Permalink
                richardm Richard McCreadie added a comment -

                Closing issue, will add note in Terrier 3.6 release issue to check java doc consistency for matching.models

                Show
                richardm Richard McCreadie added a comment - Closing issue, will add note in Terrier 3.6 release issue to check java doc consistency for matching.models

                  People

                  • Assignee:
                    craigm Craig Macdonald
                    Reporter:
                    frousseau Francois Rousseau
                  • Watchers:
                    0 Start watching this issue

                    Dates

                    • Created:
                      Updated:
                      Resolved: