Merge pull request #209 from JULIELab/aggregation_accuracy

Version 0.11.2. Fixes #207,#208.

gepi/README.md

@@ -19,8 +19,8 @@ The `production` stage expects that the complete GePI project has been built in
 Run the following commands to create a `development` container:
 
 ```bash
-DOCKER_BUILDKIT=1 docker build -t gepi:0.11.1 --target development .
-docker run -dp 8080:8080 -v {/path/to/gepi/directory}:/var/gepi/dev -e GEPI_CONFIGURATION=<path to config file> gepi:0.11.1
+DOCKER_BUILDKIT=1 docker build -t gepi:0.11.2 --target development .
+docker run -dp 8080:8080 -v {/path/to/gepi/directory}:/var/gepi/dev -e GEPI_CONFIGURATION=<path to config file> gepi:0.11.2
 ```
 
 The first command builds an image of the `development` stage. This will also build the `dependencies` stage where all the Java dependencies of the GePI application are downloaded and cached. This will take a while on the first execution but should be faster afterwards thanks to caching.
@@ -38,8 +38,8 @@ To run the `production` container, run
 
 ```bash
 mvn clean package --projects gepi-webapp --also-make
-DOCKER_BUILDKIT=1 docker build -t gepi:0.11.1 --target production .
-docker run -dp 8080:8080 --name gepi gepi:0.11.1
+DOCKER_BUILDKIT=1 docker build -t gepi:0.11.2 --target production .
+docker run -dp 8080:8080 --name gepi gepi:0.11.2
 ```
 
 These commands
@@ -68,7 +68,7 @@ gepi.neo4j.bolt.url=bolt://<host>:<port>
 
 A production environment has a few requirements that are of lesser importance during development. This section explains requirements and solutions that may come up during GePI deployment with the Docker container. While detailed explanations come below, the full Docker `run` command we use for deployment looks like the following:
 ```
-docker run -dp 80:8080 -p 443:8443 -v /host/path/to/certificate.p12:/var/lib/jetty/etc/keystore.p12 -v /host/path/to/configuration.properties:/gepi-webapp-configuration.properties --add-host=host.docker.internal:host-gateway --name gepi -e GEPI_CONFIGURATION=/gepi-webapp-configuration.properties gepi:0.11.1 jetty.sslContext.keyStorePassword=<changeit>
+docker run -dp 80:8080 -p 443:8443 -v /host/path/to/certificate.p12:/var/lib/jetty/etc/keystore.p12 -v /host/path/to/configuration.properties:/gepi-webapp-configuration.properties --add-host=host.docker.internal:host-gateway --name gepi -e GEPI_CONFIGURATION=/gepi-webapp-configuration.properties gepi:0.11.2 jetty.sslContext.keyStorePassword=<changeit>
 ```
 Alternatively, the `docker-compose-webapp.yml` file can be used with a few additions.
 
@@ -103,7 +103,8 @@ Update the new version number in the following places:
 * `pom.xml` files (tip: use `mvn versions:set -DnewVersion=<new version>` and `mvn versions:commit` to remove the backup files)
 * `README.md` (by executing `mvn clean package -DskipTests=true` to filter the `readme-raw/README.md` file to automatically set the current version to the `README.md` file)
 * `AppModule.java` in `gepi-webapp`
+ * set `PRODUCTION_MODE` to true for releases
 * the Docker image version in the `docker-compose.yml`
 * the DB version in `gene-database.xml` in the `gepi-concept-database` module
-* in `gepi-indexing-base` execute `python ../../../../jcore-misc/jcore-scripts/createMetaDescriptors.py -c -i -r manual -v 1.0 .` given that `jcore-misc` has been cloned to the same directory as GePI
+* in `gepi`-code base directory, execute `python ../../jcore-misc/jcore-scripts/createMetaDescriptors.py -c -i -r manual -v 1.0 gepi-indexing/gepi-indexing-base` given that `jcore-misc` has been cloned to the same directory as GePI
  * this updates the description file for the use with the JCoRe pipeline builder

gepi/docker-compose-webapp.yml

@@ -1,7 +1,7 @@
 version: "3.2"
 services:
  gepi:
- image: gepi:0.11.1
+ image: gepi:0.11.2
  container_name: gepi
  ports:
  - 0.0.0.0:80:8080

gepi/docker-compose-with-es.yml

@@ -1,6 +1,6 @@
 services:
  gepi:
- image: gepi:0.11.1
+ image: gepi:0.11.2
  container_name: gepi
  ports:
  - 0.0.0.0:80:8080

gepi/gepi-concept-database/pom.xml

@@ -4,7 +4,7 @@
  <parent>
  <groupId>de.julielab</groupId>
  <artifactId>gepi</artifactId>
- <version>0.11.1</version>
+ <version>0.11.2</version>
  </parent>
  <artifactId>gepi-concept-database</artifactId>
  <name>GePi Concept Database</name>

gepi/gepi-concept-database/src/main/resources/gene-database.xml

@@ -5,7 +5,7 @@
  http://www.julielab.de/conceptdb/facets/default http://www.julielab.de/conceptdb/facets/defaultfacet-1.1.0.xsd
  http://julielab.de/conceptdb/concepts/bioportal http://www.julielab.de/conceptdb/concepts/bioportalconcepts-1.1.0.xsd">
  <versioning>
- <version>0.11.1</version>
+ <version>0.11.2</version>
  </versioning>
  <connection>
  <uri>http://localhost:7474</uri>

gepi/gepi-core/pom.xml

@@ -7,7 +7,7 @@
  <parent>
  <groupId>de.julielab</groupId>
  <artifactId>gepi</artifactId>
- <version>0.11.1</version>
+ <version>0.11.2</version>
  <relativePath>../pom.xml</relativePath>
  </parent>
  <dependencies>

gepi/gepi-core/src/main/java/de/julielab/gepi/core/retrieval/data/GepiRequestData.java

@@ -18,14 +18,19 @@ public class GepiRequestData implements Cloneable {
  private String filterFieldsConnectionOperator = "AND";
  private EnumSet<InputMode> inputMode;
  private String docId;
+ private int eventRetrievalLimitForAggregations;
  private long dataSessionId;
  private boolean includeUnary;
  private int eventLikelihood;
  private String[] taxId;
  private String sectionNameFilterString;
  private int pageSize = 10;
 
- public GepiRequestData(List<String> eventTypes, boolean includeUnary, int eventLikelihood, Future<IdConversionResult> listAGePiIds, Future<IdConversionResult> listBGePiIds, String[] taxId, String sentenceFilterString, String paragraphFilterString, String filterFieldsConnectionOperator, String sectionNameFilterString, EnumSet<InputMode> inputMode, String docId, long dataSessionId) {
+ public int getEventRetrievalLimitForAggregations() {
+ return eventRetrievalLimitForAggregations;
+ }
+
+ public GepiRequestData(List<String> eventTypes, boolean includeUnary, int eventLikelihood, Future<IdConversionResult> listAGePiIds, Future<IdConversionResult> listBGePiIds, String[] taxId, String sentenceFilterString, String paragraphFilterString, String filterFieldsConnectionOperator, String sectionNameFilterString, EnumSet<InputMode> inputMode, String docId, int eventRetrievalLimitForAggregations, long dataSessionId) {
  this.includeUnary = includeUnary;
  this.eventLikelihood = eventLikelihood;
  this.taxId = taxId;
@@ -38,6 +43,7 @@ public GepiRequestData(List<String> eventTypes, boolean includeUnary, int eventL
  this.filterFieldsConnectionOperator = filterFieldsConnectionOperator;
  this.inputMode = inputMode;
  this.docId = docId;
+ this.eventRetrievalLimitForAggregations = eventRetrievalLimitForAggregations;
  this.dataSessionId = dataSessionId;
  }
 

gepi/gepi-core/src/main/java/de/julielab/gepi/core/retrieval/services/EventRetrievalService.java

@@ -147,7 +147,7 @@ public class EventRetrievalService implements IEventRetrievalService {
  FIELD_EVENT_ARG_HOMOLOGY_PREFERRED_NAME,
  FIELD_NUM_ARGUMENTS
  );
- private static final int SCROLL_SIZE = 200;
+ private static final int SCROLL_SIZE = 2000;
  private Logger log;
  private ISearchServerComponent searchServerComponent;
  private String documentIndex;
@@ -218,8 +218,8 @@ public CompletableFuture<EventRetrievalResult> closedSearch(GepiRequestData requ
  serverRqst.index = documentIndex;
  serverRqst.start = from;
  serverRqst.rows = numRows;
- serverRqst.requestTimeout = "10m";
- configureDeepPaging(serverRqst, downloadAll, forCharts);
+// serverRqst.requestTimeout = "10m";
+ configureDeepPaging(serverRqst, downloadAll, forCharts, requestData.getEventRetrievalLimitForAggregations());
  if (!downloadAll) {
  addHighlighting(serverRqst);
  }
@@ -230,7 +230,7 @@ public CompletableFuture<EventRetrievalResult> closedSearch(GepiRequestData requ
  log.debug("Sent closed search server request");
  searchServerComponent.process(carrier);
  if (log.isDebugEnabled())
- log.debug("Server answered after {} seconds. Reading results.", (System.currentTimeMillis()-time) / 1000);
+ log.debug("Server answered after {} seconds. Reading results.", (System.currentTimeMillis() - time) / 1000);
 
  EventRetrievalResult eventResult = eventResponseProcessingService
  .getEventRetrievalResult(carrier.getSingleSearchServerResponse());
@@ -303,7 +303,7 @@ public CompletableFuture<EventRetrievalResult> openSearch(GepiRequestData gepiRe
  log.debug("Sent open search server request");
  searchServerComponent.process(carrier);
  if (log.isDebugEnabled())
- log.debug("Server answered after {} seconds. Reading results.", (System.currentTimeMillis()-time) / 1000);
+  log.debug("Server answered after {} seconds. Reading results.", (System.currentTimeMillis() - time) / 1000);
 
 
  EventRetrievalResult eventResult = eventResponseProcessingService
@@ -342,23 +342,24 @@ public SearchServerRequest getOpenSearchRequest(GepiRequestData requestData, int
  serverRqst.index = documentIndex;
  serverRqst.start = from;
  serverRqst.rows = numRows;
- serverRqst.requestTimeout = "10m";
- configureDeepPaging(serverRqst, downloadAll, forCharts);
+// serverRqst.requestTimeout = "10m";
+ configureDeepPaging(serverRqst, downloadAll, forCharts, requestData.getEventRetrievalLimitForAggregations());
  if (!downloadAll) {
  addHighlighting(serverRqst);
  }
  return serverRqst;
  }
 
- private void configureDeepPaging(SearchServerRequest serverRqst, boolean downloadAll, boolean forCharts) {
+ private void configureDeepPaging(SearchServerRequest serverRqst, boolean downloadAll, boolean forCharts, int interactionRetrievalLimit) {
  if (downloadAll)
- serverRqst.rows = SCROLL_SIZE;
+ serverRqst.rows = Math.min(SCROLL_SIZE, interactionRetrievalLimit);
  serverRqst.fieldsToReturn = forCharts ? FIELDS_FOR_CHARTS : FIELDS_FOR_TABLE;
- serverRqst.downloadCompleteResults = downloadAll;
+ serverRqst.downloadCompleteResults = downloadAll && interactionRetrievalLimit > 0;
  serverRqst.downloadCompleteResultsMethod = "searchAfter";
  serverRqst.downloadCompleteResultMethodKeepAlive = "5m";
  if (downloadAll) {
- serverRqst.downloadCompleteResultsLimit = 200;
+ if (interactionRetrievalLimit < Integer.MAX_VALUE)
+ serverRqst.downloadCompleteResultsLimit = interactionRetrievalLimit;
  serverRqst.addSortCommand("_shard_doc", SortOrder.ASCENDING);
  }
  }
@@ -424,8 +425,8 @@ public CompletableFuture<EventRetrievalResult> getFulltextFilteredEvents(GepiReq
  serverRqst.index = documentIndex;
  serverRqst.start = from;
  serverRqst.rows = numRows;
- serverRqst.requestTimeout = "10m";
- configureDeepPaging(serverRqst, downloadAll, forCharts);
+// serverRqst.requestTimeout = "10m";
+ configureDeepPaging(serverRqst, downloadAll, forCharts, requestData.getEventRetrievalLimitForAggregations());
  if (!downloadAll) {
  addHighlighting(serverRqst);
  }
@@ -436,7 +437,7 @@ public CompletableFuture<EventRetrievalResult> getFulltextFilteredEvents(GepiReq
  log.debug("Sent full-text search server request");
  searchServerComponent.process(carrier);
  if (log.isDebugEnabled())
- log.debug("Server answered after {} seconds. Reading results.", (System.currentTimeMillis()-time) / 1000);
+ log.debug("Server answered after {} seconds. Reading results.", (System.currentTimeMillis() - time) / 1000);
 
  EventRetrievalResult eventResult = eventResponseProcessingService
  .getEventRetrievalResult(carrier.getSingleSearchServerResponse());

gepi/gepi-indexing/gepi-indexing-base/component.meta

@@ -18,7 +18,7 @@
  "maven-artifact": {
  "artifactId": "gepi-indexing-base",
  "groupId": "de.julielab",
- "version": "0.11.1"
+ "version": "0.11.2"
  },
  "name": "GePi Indexing Base"
 }

gepi/gepi-indexing/gepi-indexing-base/pom.xml

@@ -5,7 +5,7 @@
  <parent>
  <artifactId>gepi-indexing</artifactId>
  <groupId>de.julielab</groupId>
- <version>0.11.1</version>
+ <version>0.11.2</version>
  </parent>
  <modelVersion>4.0.0</modelVersion>
 

gepi/gepi-indexing/gepi-indexing-base/src/main/java/de/julielab/gepi/indexing/RelationFieldValueGenerator.java

@@ -194,6 +194,7 @@ else if (j > i)
 // document.addField("argument1prefname", createRawFieldValueForAnnotation(argPair[0], arg1EntryIdPath, geneFb.egid2prefNameReplaceFilter));
  final IFieldValue arg1HomoPrefNameValue = createRawFieldValueForAnnotation(argPair[0], arg1EntryIdPath, geneFb.orgid2topaggprefname);
 // document.addField("argument1homoprefname", arg1HomoPrefNameValue);
+ document.addField("argument1homoprefnameaggvalue", arg1HomoPrefNameValue);
 // final IFieldValue arg1GoPrefnames = createRawFieldValueForAnnotation(argPair[0], arg1EntryIdPath, geneFb.eg2goprefnameFilter);
 // document.addField("argument1goprefnames", arg1GoPrefnames);
 // document.addField("argument1matchtype", Stream.of(argPair).map(ArgumentMention.class::cast).map(ArgumentMention::getRef).map(ConceptMention.class::cast).map(cm -> cm.getResourceEntryList(0).getConfidence() == null || cm.getResourceEntryList(0).getConfidence().contains("9999") ? "exact" : "fuzzy").toArray());
@@ -211,6 +212,7 @@ else if (j > i)
 // document.addField("argument2prefname", createRawFieldValueForFieldValue(document.getAsRawToken("argument2conceptid"), geneFb.conceptid2prefNameFilter));
  final IFieldValue arg2HomoPrefNameValue = createRawFieldValueForAnnotation(argPair[1], arg2EntryIdPath, geneFb.orgid2topaggprefname);
 // document.addField("argument2homoprefname", arg2HomoPrefNameValue);
+ document.addField("argument2homoprefnameaggvalue", arg2HomoPrefNameValue);
 // final IFieldValue arg2GoPrefnames = createRawFieldValueForAnnotation(argPair[1], arg2EntryIdPath, geneFb.eg2goprefnameFilter);
 // document.addField("argument2goprefnames", arg2GoPrefnames);
 // document.addField("argument2matchtype", Stream.of(argPair).map(ArgumentMention.class::cast).map(ArgumentMention::getRef).map(ConceptMention.class::cast).map(cm -> cm.getResourceEntryList(0).getConfidence() == null || cm.getResourceEntryList(0).getConfidence().contains("9999") ? "exact" : "fuzzy").toArray());
@@ -255,7 +257,7 @@ else if (j > i)
  document.addField("ARGUMENT_FS", argPair);
  // For ElasticSearch aggregations, we create terms in the form 'symbol1---symbol2'. We also sort the symbols so that the same pair of symbols is always stored in the same order.
  // Then we can use ElasticSearch aggregations to count interactions occurrences instead of retrieving all documents and counting ourselves.
-// document.addField("aggregationvalue", document.getAsArrayFieldValue("argumenthomoprefnames").stream().map(IFieldValue::toString).sorted().collect(Collectors.joining("---")));
+ document.addField("aggregationvalue", document.getAsArrayFieldValue("argumenthomoprefnames").stream().map(IFieldValue::toString).sorted().collect(Collectors.joining("---")));
 // final ArrayFieldValue go1Values = new ArrayFieldValue(arg1GoPrefnames);
 // final ArrayFieldValue go2Values = new ArrayFieldValue(arg2GoPrefnames);
 // for (IFieldValue go1 : go1Values) {

gepi/gepi-indexing/gepi-indexing-base/src/main/resources/elasticSearchMapping.json

@@ -103,6 +103,9 @@
  "store": true,
  "norms": false
  },
+ "argument1homoprefnameaggvalue": {
+ "type": "keyword"
+ },
  "argument1conceptid": {
  "type": "keyword",
  "store": true
@@ -158,6 +161,9 @@
  "store": true,
  "norms": false
  },
+ "argument2homoprefnameaggvalue": {
+ "type": "keyword"
+ },
  "argument2conceptid": {
  "type": "keyword",
  "store": true

gepi/gepi-indexing/gepi-indexing-debug/pom.xml

@@ -5,7 +5,7 @@
  <parent>
  <artifactId>gepi-indexing</artifactId>
  <groupId>de.julielab</groupId>
- <version>0.11.1</version>
+ <version>0.11.2</version>
  <relativePath>../pom.xml</relativePath>
  </parent>
  <modelVersion>4.0.0</modelVersion>
@@ -15,7 +15,7 @@
  <dependency>
  <groupId>de.julielab</groupId>
  <artifactId>gepi-indexing-base</artifactId>
- <version>0.11.1</version>
+ <version>0.11.2</version>
  </dependency>
  </dependencies>
 </project>