Merge pull request #70 from beclab/feat/similarity-experiment

fix: not print highest score when there is no valid algorithm item

Author: bleachzou3

train-rank/src/knowledgebase_api.cpp

@@ -1125,6 +1125,7 @@ namespace knowledgebase
     http_request req(methods::GET);
     req.set_request_uri(U(builder.to_string()));
     std::string blf_user_env_value = getEnvString(BFL_USER_ENV_NAME, "");
+    LOG(DEBUG) << "blf_user_env_value " << blf_user_env_value << std::endl;
     req.headers()
         .add(U(X_BFL_USER_HEADER), U(blf_user_env_value));
 
@@ -1148,9 +1149,12 @@ namespace knowledgebase
           }
           LOG(DEBUG) << "code " << code << " message " << message << std::endl;
           if (v.has_string_field("data")) {
-            std::string last_rankt_time_data = v.at("data").as_string();
-            convertStringTimestampToInt64(last_rankt_time_data,
+            std::string last_extractor_time_data = v.at("data").as_string();
+            LOG(DEBUG) << "last_extractor_time_data_str " << last_extractor_time_data
+                       << std::endl;
+            convertStringTimestampToInt64(last_extractor_time_data,
                                           &last_extractor_time);
+            LOG(DEBUG) << "compelete last_extractor_time " << last_extractor_time << std::endl;
           }
         } catch (http_exception const &e) {
           LOG(ERROR) << "Error exception " << e.what() << std::endl;

train-rank/src/rssrank.cpp

@@ -1038,6 +1038,7 @@ namespace rssrank
     int embedding_dimension = getCurrentEmbeddingDimension();
     std::vector<double> recall_user_embedding = knowledgebase::init_user_embedding(embedding_dimension);
     knowledgebase::updateRecallUserEmbedding(current_srouce_name, recall_user_embedding, current_rank_time);
+
     for (const auto &current_item : not_impressioned_algorithm_to_entry)
     {
       std::optional<Entry> temp_entry =
@@ -1080,15 +1081,24 @@ namespace rssrank
 
     std::sort(algorithm_integer_id_to_score.begin(), algorithm_integer_id_to_score.end(), [](const std::pair<int, double> &a, const std::pair<int, double> &b)
               { return a.second > b.second; });
+
+    if (id_to_score_with_meta.size() == 0)
+    {
+      LOG(ERROR) << "rankByTimeForColdStart no algorithm to rank" << std::endl;
+    }
     if (FLAGS_verbose)
     {
       std::vector<std::pair<std::string, float>> sorted_algorithm_to_score = knowledgebase::rankScoreMetadata(id_to_score_with_meta);
       for (const auto &pr : sorted_algorithm_to_score)
       {
         LOG(INFO) << "Algorithm [" << pr.first << "] score [" << pr.second << "]" << std::endl;
       }
-      std::cout << "max " << sorted_algorithm_to_score[0].second << " min " << sorted_algorithm_to_score[sorted_algorithm_to_score.size() - 1].second << std::endl;
-    }
+      if (sorted_algorithm_to_score.size() > 0)
+      {
+        std::cout << "Algorithm [" << sorted_algorithm_to_score[0].first << "] max score [" << sorted_algorithm_to_score[0].second << "]" << std::endl;
+      }
+        }
+
     if (FLAGS_upload_score)
     {
       LOG(INFO) << "score with metadata update to knowledge " << std::endl;

train-rank/test/knowledge_api_test.cpp

@@ -189,7 +189,7 @@ TEST(KnowledgeApiTest, TestGetLastExtractTime)
   // --gtest_filter=KnowledgeApiTest.TestGetLastExtractTime
   initDevelop();
   init_log();
-  int64_t last_extractor_time = knowledgebase::getLastExtractorTime("_r4business");
+  int64_t last_extractor_time = knowledgebase::getLastExtractorTime("r4business");
   std::cout << "last_extractor_time " << last_extractor_time << std::endl;
 }