Changing queries to use the esci data better. Removing unneeded data …

…scripts.

data/esci/.gitattributes

@@ -1 +1,2 @@
 ubi_queries_events_1000.ndjson.bz2 filter=lfs diff=lfs merge=lfs -text
+ubi_queries_events.ndjson.bz2 filter=lfs diff=lfs merge=lfs -text

data/esci/index.sh

@@ -1,3 +1,3 @@
 #!/bin/bash -e
 
-curl -s -X POST "http://localhost:9200/_bulk?pretty" -H "Content-Type: application/x-ndjson" --data-binary @ubi_queries_events_1000.ndjson
+curl -s -X POST "http://localhost:9200/_bulk?pretty" -H "Content-Type: application/x-ndjson" --data-binary @ubi_queries_events.ndjson

data/ubi-chorus-data-generator/README.md

@@ -0,0 +1,3 @@
+# UBI Data Generator for Chorus using the Ecommerce Index
+
+You should likely use the `esci` dataset instead. It's up one directory level.

opensearch-search-quality-evaluation-plugin/docker-compose.yaml

@@ -23,24 +23,24 @@ services:
       - "9600:9600"
     networks:
       - opensearch-net
-#    volumes:
-#      - opensearch-data1:/usr/share/opensearch/data
+    volumes:
+      - opensearch-data1:/usr/share/opensearch/data
 
-#  opensearch_sef_dashboards:
-#    image: opensearchproject/opensearch-dashboards:2.16.0
-#    container_name: opensearch_sef_dashboards
-#    ports:
-#      - "5601:5601"
-#    environment:
-#      OPENSEARCH_HOSTS: '["http://opensearch:9200"]'
-#      DISABLE_SECURITY_DASHBOARDS_PLUGIN: "true"
-#    depends_on:
-#      - opensearch_sef
-#    networks:
-#      - opensearch-net
+  opensearch_sef_dashboards:
+    image: opensearchproject/opensearch-dashboards:2.18.0
+    container_name: opensearch_sef_dashboards
+    ports:
+      - "5601:5601"
+    environment:
+      OPENSEARCH_HOSTS: '["http://opensearch_sef:9200"]'
+      DISABLE_SECURITY_DASHBOARDS_PLUGIN: "true"
+    depends_on:
+      - opensearch_sef
+    networks:
+      - opensearch-net
 
-#volumes:
-#  opensearch-data1:
+volumes:
+  opensearch-data1:
 
 networks:
   opensearch-net:

opensearch-search-quality-evaluation-plugin/src/main/java/org/opensearch/eval/judgments/clickmodel/coec/CoecClickModel.java

@@ -124,7 +124,7 @@ public long calculateCoec(final Map<Integer, Double> rankAggregatedClickThrough,
                 for(int r = 0; r < maxRank; r++) {
 
                     final double meanCtrAtRank = rankAggregatedClickThrough.getOrDefault(r, 0.0);
-                    final int countOfTimesShownAtRank = openSearchHelper.getCountOfQueriesForUserQueryHavingResultInRankR(userQuery, ctr.getObjectId(), r);
+                    final long countOfTimesShownAtRank = openSearchHelper.getCountOfQueriesForUserQueryHavingResultInRankR(userQuery, ctr.getObjectId(), r);
 
 //                    System.out.println("rank = " + r);
 //                    System.out.println("\tmeanCtrAtRank = " + meanCtrAtRank);
@@ -179,7 +179,7 @@ private Map<String, Set<ClickthroughRate>> getClickthroughRate(final int maxRank
         // TODO: Use maxRank in place of the hardcoded 20.
         // TODO: Allow for a time period and for a specific application.
 
-        /**
+        /*
          * {
          *                 "bool": {
          *                   "should": [
@@ -234,7 +234,7 @@ private Map<String, Set<ClickthroughRate>> getClickthroughRate(final int maxRank
                 "              }";
 
         final BoolQueryBuilder queryBuilder = new BoolQueryBuilder().must(new WrapperQueryBuilder(query));
-        final SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder().query(queryBuilder).size(1000);
+        final SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder().query(queryBuilder).size(500);
         final Scroll scroll = new Scroll(TimeValue.timeValueMinutes(10L));
 
         final SearchRequest searchRequest = Requests
@@ -243,6 +243,7 @@ private Map<String, Set<ClickthroughRate>> getClickthroughRate(final int maxRank
                 .scroll(scroll);
 
         SearchResponse searchResponse = client.search(searchRequest).get();
+
         String scrollId = searchResponse.getScrollId();
         SearchHit[] searchHits = searchResponse.getHits().getHits();
 
@@ -254,6 +255,8 @@ private Map<String, Set<ClickthroughRate>> getClickthroughRate(final int maxRank
 
                 final UbiEvent ubiEvent = AccessController.doPrivileged((PrivilegedAction<UbiEvent>) () -> gson.fromJson(hit.getSourceAsString(), UbiEvent.class));
 
+                //LOGGER.info("event: {}", ubiEvent.toString());
+
                 // We need to the hash of the query_id because two users can both search
                 // for "computer" and those searches will have different query IDs, but they are the same search.
                 final String userQuery = openSearchHelper.getUserQuery(ubiEvent.getQueryId());
@@ -280,7 +283,12 @@ private Map<String, Set<ClickthroughRate>> getClickthroughRate(final int maxRank
             final SearchScrollRequest scrollRequest = new SearchScrollRequest(scrollId);
             scrollRequest.scroll(scroll);
 
+            //LOGGER.info("Doing scroll to next results");
+            // TODO: Getting a warning in the log that "QueryGroup _id can't be null, It should be set before accessing it. This is abnormal behaviour"
+            // https://github.com/opensearch-project/OpenSearch/blob/f105e4eb2ede1556b5dd3c743bea1ab9686ebccf/server/src/main/java/org/opensearch/wlm/QueryGroupTask.java#L73
             searchResponse = client.searchScroll(scrollRequest).get();
+            //LOGGER.info("Scroll complete.");
+
             scrollId = searchResponse.getScrollId();
 
             searchHits = searchResponse.getHits().getHits();

opensearch-search-quality-evaluation-plugin/src/main/java/org/opensearch/eval/judgments/model/ubi/event/EventObject.java

@@ -18,6 +18,11 @@ public class EventObject {
     @SerializedName("object_id")
     private String objectId;
 
+    @Override
+    public String toString() {
+        return "[" + objectIdField + ", " + objectId + "]";
+    }
+
     public String getObjectId() {
         return objectId;
     }

opensearch-search-quality-evaluation-plugin/src/main/java/org/opensearch/eval/judgments/model/ubi/event/Position.java

@@ -15,6 +15,11 @@ public class Position {
     @SerializedName("index")
     private int index;
 
+    @Override
+    public String toString() {
+        return String.valueOf(index);
+    }
+
     public int getIndex() {
         return index;
     }

opensearch-search-quality-evaluation-plugin/src/main/java/org/opensearch/eval/judgments/model/ubi/event/UbiEvent.java

@@ -29,7 +29,7 @@ public class UbiEvent {
 
     @Override
     public String toString() {
-        return actionName + ", " + clientId + ", " + queryId + ", " + eventAttributes.getObject() + ", " + eventAttributes.getPosition().getIndex();
+        return actionName + ", " + clientId + ", " + queryId + ", " + eventAttributes.getObject().toString() + ", " + eventAttributes.getPosition().getIndex();
     }
 
     public String getActionName() {