From 1941c2a45df3f0b854424020b0220f502c6498de Mon Sep 17 00:00:00 2001
From: Alex Skrenchuk <alexskr@stanford.edu>
Date: Fri, 15 Nov 2024 15:20:42 -0800
Subject: [PATCH] remove recommender_v1

---
 lib/ncbo_annotator.rb    |  1 -
 lib/ncbo_recommender.rb  | 88 ----------------------------------------
 lib/recommendation.rb    | 35 ----------------
 test/test_recommender.rb | 47 ---------------------
 4 files changed, 171 deletions(-)
 delete mode 100644 lib/ncbo_recommender.rb
 delete mode 100644 lib/recommendation.rb
 delete mode 100644 test/test_recommender.rb

diff --git a/lib/ncbo_annotator.rb b/lib/ncbo_annotator.rb
index e1e76c3b..3fbbae28 100644
--- a/lib/ncbo_annotator.rb
+++ b/lib/ncbo_annotator.rb
@@ -14,7 +14,6 @@
 require_relative 'ncbo_annotator/mgrep/mgrep'
 require_relative 'ncbo_annotator/config'
 require_relative 'ncbo_annotator/monkeypatches'
-require_relative 'ncbo_recommender'
 
 # Require all models
 project_root = File.dirname(File.absolute_path(__FILE__))
diff --git a/lib/ncbo_recommender.rb b/lib/ncbo_recommender.rb
deleted file mode 100644
index 03e72d74..00000000
--- a/lib/ncbo_recommender.rb
+++ /dev/null
@@ -1,88 +0,0 @@
-require 'logger'
-require 'ontologies_linked_data'
-require_relative 'recommendation'
-
-module Recommender
-  module Models
-
-    class NcboRecommender
-
-      def initialize()
-        @logger = Kernel.const_defined?("LOGGER") ? Kernel.const_get("LOGGER") : Logger.new(STDOUT)
-      end
-
-      DEFAULT_HIERARCHY_LEVELS = 5
-
-      def recommend(text, ontologies=[], include_classes=false)
-        annotator = Annotator::Models::NcboAnnotator.new
-        annotations = annotator.annotate(text, {
-            ontologies: ontologies,
-            semantic_types: [],
-            filter_integers: false,
-            expand_class_hierarchy: true,
-            expand_hierarchy_levels: DEFAULT_HIERARCHY_LEVELS,
-            expand_with_mappings: false,
-            min_term_size: nil,
-            whole_word_only: true,
-            with_synonyms: true
-        })
-
-        recommendations = {}
-        classes_matched = []
-
-        annotations.each do |ann|
-          classId = ann.annotatedClass.id.to_s
-          ont = ann.annotatedClass.submission.ontology
-          ontologyId = ont.id.to_s
-
-          unless recommendations.include?(ontologyId)
-            cls_count = get_ontology_class_count(ont)
-            next if cls_count <= 0  # skip any ontologies without a ready latest submission
-            recommendations[ontologyId] = Recommendation.new
-            recommendations[ontologyId].ontology = ont
-            recommendations[ontologyId].numTermsTotal = cls_count
-          end
-
-          rec = recommendations[ontologyId]
-          cls_ont_key = "#{classId}_#{ontologyId}"
-          unless classes_matched.include?(cls_ont_key)
-            classes_matched << cls_ont_key
-            rec.annotatedClasses << ann.annotatedClass if include_classes
-            rec.numTermsMatched += 1
-          end
-          rec.increment_score(ann)
-        end
-
-        recommendations.values.each {|v| v.normalize_score}
-        return recommendations.values.sort {|a,b| b.score <=> a.score}
-      end
-
-
-      private
-
-      def get_ontology_class_count(ont)
-        sub = nil
-        begin
-          #TODO: there appears to be a bug that does not allow retrieving submission by its id
-          # because the id is incorrect. The workaround is to get the ontology object and
-          # then retrieve its latest submission.
-          sub = LinkedData::Models::Ontology.find(ont.id).first.latest_submission
-        rescue
-          @logger.error("Unable to retrieve latest submission for #{ont.id.to_s} in Recommender.")
-        end
-        return 0 if sub.nil?
-        begin
-          sub.bring(metrics: LinkedData::Models::Metric.attributes)
-          cls_count = sub.metrics.classes
-        rescue
-          @logger.error("Unable to retrieve metrics for latest submission of #{ont.id.to_s} in Recommender.")
-          cls_count = LinkedData::Models::Class.where.in(sub).count
-        end
-        return cls_count || 0
-      end
-
-
-    end
-
-  end
-end
diff --git a/lib/recommendation.rb b/lib/recommendation.rb
deleted file mode 100644
index 3ced1ded..00000000
--- a/lib/recommendation.rb
+++ /dev/null
@@ -1,35 +0,0 @@
-module Recommender
-
-  class Recommendation
-
-    include LinkedData::Hypermedia::Resource
-
-    attr_accessor :ontology, :score, :numTermsMatched, :numTermsTotal, :annotatedClasses
-
-    embed :annotatedClasses
-    
-    def initialize
-      @score = 0
-      @numTermsMatched = 0
-      @numTermsTotal = 0
-      @annotatedClasses = []
-    end
-
-    def increment_score(annotation)
-      annotation.annotations.each do |occ|
-        if occ[:matchType] == "PREF"
-          @score += 10
-        elsif occ[:matchType] == "SYN"
-          @score += 5
-        end
-      end
-      @score += annotation.hierarchy.length * 2
-    end
-
-    def normalize_score()
-      @score = (@score / Math.log10(@numTermsTotal)).round(2)
-    end
-
-  end
-
-end
\ No newline at end of file
diff --git a/test/test_recommender.rb b/test/test_recommender.rb
deleted file mode 100644
index b3c4bca3..00000000
--- a/test/test_recommender.rb
+++ /dev/null
@@ -1,47 +0,0 @@
-require_relative 'test_case'
-require 'json'
-require 'redis'
-
-class TestRecommender < TestCase
-
-  def self.before_suite
-    @@text = <<eos
-Ginsenosides chemistry, biosynthesis, analysis, and potential health effects in software concept or data." "Ginsenosides are a special group of triterpenoid saponins that can be classified into two groups by the skeleton of their aglycones, namely dammarane- and oleanane-type. Ginsenosides are found nearly exclusively in Panax species (ginseng) and up to now more than 150 naturally occurring ginsenosides have been isolated from roots, leaves/stems, fruits, and/or flower heads of ginseng. The same concept indicates Ginsenosides have been the target of a lot of research as they are believed to be the main active principles behind the claims of ginsengs efficacy. The potential health effects of ginsenosides that are discussed in this chapter include anticarcinogenic, immunomodulatory, anti-inflammatory, antiallergic, antiatherosclerotic, antihypertensive, and antidiabetic effects as well as antistress activity and effects on the central nervous system. Ginsensoides can be metabolized in the stomach (acid hydrolysis) and in the gastrointestinal tract (bacterial hydrolysis) or transformed to other ginsenosides by drying and steaming of ginseng to more bioavailable and bioactive ginsenosides. The metabolization and transformation of intact ginsenosides, which seems to play an important role for their potential health effects, are discussed. Qualitative and quantitative analytical techniques for the analysis of ginsenosides are important in relation to quality control of ginseng products and plant material and for the determination of the effects of processing of plant material as well as for the determination of the metabolism and bioavailability of ginsenosides. Analytical techniques for the analysis of ginsenosides that are described in this chapter are thin-layer chromatography (TLC), high-performance liquid chromatography (HPLC) combined with various detectors, gas chromatography (GC), colorimetry, enzyme immunoassays (EIA), capillary electrophoresis (CE), nuclear magnetic resonance (NMR) spectroscopy, and spectrophotometric methods.
-eos
-    @@recommender = Recommender::Models::NcboRecommender.new
-  end
-
-  def self.after_suite
-  end
-
-  def test_recommend
-    recommendations = @@recommender.recommend(@@text)
-    assert_equal(3, recommendations.length, msg='Failed to return 3 recommendations')
-  end
-
-  def test_recommend_with_ontologies
-    ont_acronyms = ["BROTEST-0", "MCCLTEST-0"]
-    include_classes = false # same as default
-    recommendations = @@recommender.recommend(@@text, ont_acronyms, include_classes)
-    assert_equal(2, recommendations.length)
-    recommendations.each do |rec|
-      assert_instance_of(Recommender::Recommendation, rec, msg='recommendations value is not an instance of Recommender::Recommendation')
-      assert_equal(0, rec.annotatedClasses.length, msg='Returned annotatedClasses when they were not requested' )
-    end
-    rec_acronyms = recommendations.map {|rec| rec.ontology.acronym }
-    assert_equal(ont_acronyms, rec_acronyms, msg="#{rec_acronyms} != #{ont_acronyms}")
-  end
-
-  def test_recommend_with_classes
-    ont_acronyms = []  # no restrictions
-    include_classes = true
-    recommendations = @@recommender.recommend(@@text, ont_acronyms, include_classes)
-    assert_equal(3, recommendations.length)
-    recommendations.each do |rec|
-      assert_instance_of(Recommender::Recommendation, rec, msg='recommendations value is not an instance of Recommender::Recommendation')
-      assert(rec.annotatedClasses.length > 0, msg='Failed to return annotatedClasses')
-    end
-  end
-
-end
-