uc3-dmp-id 0.1.35 → 0.1.37

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5828afe1afaf7214cdea0e02ed8b834587068be56f1f1a060050c7768c4d19a8
-  data.tar.gz: 10c0bb60320d71414959a9623e435d71ab52b74b758a38fc23cb7fad68f5bd3e
+  metadata.gz: 4b8c0ff9aa6add1fc2da937d9f76872ae6bb287adf5df008d9c0d434f25fd717
+  data.tar.gz: a0479d353eb1d64cfab1bdb718f3efcf6a1d66cec8ea444ab735e74708f88984
 SHA512:
-  metadata.gz: f4195fb80461a995b42d10e41ff6828aa79692d3cf92c7ce942f71f830c74d2544806a4c0ada2924d0aa7bf8c310c8b8f03a63eeacbd8754d4262aeb343f7dba
-  data.tar.gz: a1108d2311d8c8331b2b176278ae10d9283bdcbb0f4568c0b7d488e22268293b0356c36a51937857f709c64be89edcf2c8a8f98d74d98a66eda3c08f9ef0dea2
+  metadata.gz: a578c6b89bb5d94172d5e2c24255f81326871616767039a4aa4de6d1c00da4a9d16fcf20f26a4479d8d1d5017e90dfcb27ca9c6c4bb5b1ed5421928bfb518eeb
+  data.tar.gz: 3dc8d1999fb6f26cdf8fc465d160513982ac7607bfb7c5e4c5f719b4f43d3b5d82261c5c4ce87cdd10be1d1631276b212b54cfd60b3f5523b389b3f4da25dbe5

data/lib/uc3-dmp-id/comparator.rb

@@ -24,6 +24,8 @@ module Uc3DmpId
       @details_hash = {}
 
       @dmps = args.fetch(:dmps, [])
+
+      @logger&.debug(message: 'Comparator DMPs', details: @dmps)
       raise ComparatorError, MSG_MISSING_DMPS if @dmps.empty?
     end
 
@@ -57,11 +59,12 @@ module Uc3DmpId
       return scoring unless hash.is_a?(Hash) && !hash['title'].nil?
 
       @dmps.each do |dmp|
+        @logger&.debug(message: 'Incoming external work', details: hash)
         # Compare the grant ids. If we have a match return the response immediately since that is
         # a very positive match!
         response = { confidence: 'None', score: 0, notes: [] }
         response = _grants_match?(array: hash.fetch('grant_ids', []), dmp:, response:)
-        scoring << respoonse if response[:confidence] != 'None'
+        scoring << response if response[:confidence] != 'None'
         next if response[:confidence] != 'None'
 
         # Compare the people involved, their affiliations and any funding opportunity numbers
@@ -73,8 +76,8 @@ module Uc3DmpId
         response = _repository_match?(hash:, dmp:, response:) if response[:score].positive?
         response = _text_match?(type: 'title', text: hash['title'], dmp:, response:) if response[:score].positive?
         response = _text_match?(type: 'abstract', text: hash['description'], dmp:, response:) if response[:score].positive?
-        # If the score is less than 3 then we have no confidence that it is a match
-        next if response[:score] <= 2
+        # If the score is less than 4 then we have no confidence that it is a match
+        next if response[:score] <= 3
 
         # Set the confidence level based on the score
         response[:dmp_id] = "DMP##{dmp['dmp_id']}"
@@ -110,6 +113,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['grant_ids'], array_b: ids)
       return response if matched <= 0
 
+      @logger&.debug(message: 'Grant ID match!', details: { dmp: dmp['grant_ids'], work: ids })
       response[:confidence] = 'Absolute'
       response[:score] = 100
       response[:notes] << 'the grant ID matched'
@@ -133,6 +137,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['funder_opportunity_ids'], array_b: ids)
       return response if matched <= 0
 
+      @logger&.debug(message: 'Opportunities match!', details: { dmp: dmp['funder_opportunity_ids'], work: ids })
       response[:score] += 5
       response[:notes] << 'the funding opportunity number matched'
       response
@@ -159,6 +164,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['people_ids'], array_b: ids)
       return response if matched <= 0
 
+      @logger&.debug(message: 'ORCID match!', details: { dmp: dmp['people_ids'], work: ids })
       response[:score] += (matched * 2)
       response[:notes] << 'contributor ORCIDs matched'
       response
@@ -182,8 +188,16 @@ module Uc3DmpId
       affil_names_matched = _compare_arrays(array_a: dmp.fetch('affiliations', []), array_b: hash['affiliations'])
       return response if last_names_matched <= 0 && rors_matched <= 0 && affil_names_matched <= 0
 
+      @logger&.debug(
+        message: 'Contributor name match',
+        details: {
+          people: { dmp: dmp['people'], work: hash['people'] },
+          rors: { dmp: dmp['affiliation_ids'], work: hash['affiliation_ids'] },
+          places: { dmp: dmp['affiliations'], work: hash['affiliations'] }
+        }
+      )
       response[:score] += last_names_matched + rors_matched + affil_names_matched
-      response[:notes] << 'contributor names and affiliations matched'
+      response[:notes] << 'contributor names and/or affiliations matched'
       response
     end
     # rubocop:enable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
@@ -201,6 +215,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['repo_ids'], array_b: hash['repo_ids'])
       return response if matched <= 0
 
+      @logger&.debug(message: 'Repos match!', details: { dmp: dmp['repo_ids'], work: hash['repo_ids'] })
       response[:score] += matched
       response[:notes] << 'repositories matched'
       response
@@ -224,6 +239,8 @@ module Uc3DmpId
       # @logger&.debug(message: 'Text::WhiteSimilarity score', details:)
       return response if details[:nlp_score] < 0.5
 
+      @logger&.debug(message: 'Titles match', details: { dmp: dmp['title'], work: text }) if type == 'title'
+      @logger&.debug(message: 'Abstracts match', details: { dmp: dmp['description'], work: text }) unless type == 'title'
       response[:score] += details[:nlp_score] >= 0.75 ? 5 : 2
       response[:notes] << "#{type}s are similar"
       response

data/lib/uc3-dmp-id/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.35'
+  VERSION = '0.1.37'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: uc3-dmp-id
 version: !ruby/object:Gem::Version
-  version: 0.1.35
+  version: 0.1.37
 platform: ruby
 authors:
 - Brian Riley
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-05-01 00:00:00.000000000 Z
+date: 2024-05-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: json