Merge pull request bst-mug#94 from michelole/issue-45

Issue 45

src/main/java/at/medunigraz/imi/bst/trec/RunnerDemo.java

@@ -27,49 +27,68 @@ public class RunnerDemo {
 	private static final Logger LOG = LogManager.getLogger();
 
 	public static void main(String[] args) {
-		final String[] runIds = { "example", "extra" };
-		final String suffix = "pmid";
-
-		final File template = new File(RunnerDemo.class.getResource("/templates/must-match-disease.json").getFile());
+		String[] pmRuns = { "example-pmid", "extra-pmid" };
+
+		final File pmTemplate = new File(RunnerDemo.class.getResource("/templates/must-match-disease.json").getFile());
 		Gene.Field[] expandTo = { Gene.Field.SYMBOL, Gene.Field.DESCRIPTION };
-		Query decorator = new WordRemovalQueryDecorator(new GeneExpanderQueryDecorator(expandTo,
-				new TemplateQueryDecorator(template, new ElasticSearchQuery())));
+		Query pmDecorator = new WordRemovalQueryDecorator(new GeneExpanderQueryDecorator(expandTo,
+				new TemplateQueryDecorator(pmTemplate, new ElasticSearchQuery("trec"))));
 
-		for (String id : runIds) {
-			LOG.info("Running collection '" + id + "'...");
-			File example = new File(CSVStatsWriter.class.getResource("/topics/" + id + ".xml").getPath());
-			TopicSet topicSet = new TopicSet(example);
+		for (String id : pmRuns) {
+			runExperiment(id, pmDecorator);
+		}
+
+
+		// TODO DRY Issue #53
+		String[] ctRuns = { "extra-ct" };
 
-			File output = new File("results/" + id + "-" + suffix + ".trec_results");
-			TrecWriter tw = new TrecWriter(output);
+		final File ctTemplate = new File(RunnerDemo.class.getResource("/templates/baseline-ct.json").getFile());
+		Query ctDecorator = new TemplateQueryDecorator(ctTemplate, new ElasticSearchQuery("clinicaltrials"));
+
+		for (String id : ctRuns) {
+			runExperiment(id, ctDecorator);
+		}
+	}
+
+	private static void runExperiment(String id, Query decorator) {
+		final String collection = id.substring(0, id.indexOf('-'));
+
+		LOG.info("Running collection '" + id + "'...");
+
+		File example = new File(CSVStatsWriter.class.getResource("/topics/" + collection + ".xml").getPath());
+		TopicSet topicSet = new TopicSet(example);
 
-			// TODO DRY Issue #53
-			Set<ResultList> resultListSet = new HashSet<>();
-			for (Topic topic : topicSet.getTopics()) {
-				List<Result> results = decorator.query(topic);
+		File output = new File("results/" + id + ".trec_results");
+		TrecWriter tw = new TrecWriter(output);
 
-				ResultList resultList = new ResultList(topic);
-				resultList.setResults(results);
-				resultListSet.add(resultList);
-			}
+		// TODO DRY Issue #53
+		Set<ResultList> resultListSet = new HashSet<>();
+		for (Topic topic : topicSet.getTopics()) {
+			List<Result> results = decorator.query(topic);
 
-			tw.write(resultListSet);
-			tw.close();
+			ResultList resultList = new ResultList(topic);
+			resultList.setResults(results);
+			resultListSet.add(resultList);
+		}
 
-			File goldStandard = new File(
-					CSVStatsWriter.class.getResource("/gold-standard/" + id + "-" + suffix + ".qrels").getPath());
-			TrecEval te = new TrecEval(goldStandard, output);
+		tw.write(resultListSet);
+		tw.close();
 
-			LOG.debug("NDCG: " + te.getNDCG());
-			LOG.trace(te.getMetricsByTopic("all"));
+		File goldStandard = new File(
+				CSVStatsWriter.class.getResource("/gold-standard/" + id + ".qrels").getPath());
+		TrecEval te = new TrecEval(goldStandard, output);
 
-			XMLStatsWriter xsw = new XMLStatsWriter(new File("stats/" + id + "-" + suffix + ".xml"));
-			xsw.write(te.getMetrics());
-			xsw.close();
+		LOG.debug("NDCG: " + te.getNDCG());
+		LOG.trace(te.getMetricsByTopic("all"));
 
-			CSVStatsWriter csw = new CSVStatsWriter(new File("stats/" + id + "-" + suffix + ".csv"));
-			csw.write(te.getMetrics());
-			csw.close();
-		}
+		XMLStatsWriter xsw = new XMLStatsWriter(new File("stats/" + id + ".xml"));
+		xsw.write(te.getMetrics());
+		xsw.close();
+
+		CSVStatsWriter csw = new CSVStatsWriter(new File("stats/" + id + ".csv"));
+		csw.write(te.getMetrics());
+		csw.close();
 	}
+
+
 }

src/main/java/at/medunigraz/imi/bst/trec/model/Result.java

@@ -11,7 +11,7 @@ public Result(String id, float score) {
 	}
 
 	public String getId() {
-		return id;
+		return id.replace("NCT", "");
 	}
 
 	public float getScore() {

src/main/java/at/medunigraz/imi/bst/trec/model/Topic.java

@@ -8,7 +8,6 @@
 import javax.xml.parsers.DocumentBuilder;
 import javax.xml.parsers.DocumentBuilderFactory;
 
-import com.sun.glass.ui.View;
 import org.w3c.dom.Document;
 import org.w3c.dom.Element;
 
@@ -162,6 +161,8 @@ public Map<String, String> getAttributes() {
 		ret.put("variant", gene);	// Backwards compatibility
 		ret.put("demographic", demographic);
 		ret.put("other", other);
+		ret.put("sex", getSex());
+		ret.put("age", String.valueOf(getAge()));
 
 		return ret;
 	}

src/main/java/at/medunigraz/imi/bst/trec/query/ElasticSearchQuery.java

@@ -12,12 +12,15 @@ public class ElasticSearchQuery implements Query {
 
 	private JSONObject jsonQuery;
 
-	public ElasticSearchQuery() {
+	private String index;
+
+	public ElasticSearchQuery(String index) {
+		this.index = index;
 	}
 
 	@Override
 	public List<Result> query(Topic topic) {
-		ElasticSearch es = new ElasticSearch();
+		ElasticSearch es = new ElasticSearch(index);
 		return es.query(jsonQuery);
 	}
 

src/main/java/at/medunigraz/imi/bst/trec/search/ElasticSearch.java

@@ -22,6 +22,16 @@ public class ElasticSearch implements SearchEngine {
 
 	private Client client = ElasticClientFactory.getClient();
 
+	private String index;
+
+	public ElasticSearch() {
+		this.index = "_all";
+	}
+
+	public ElasticSearch(String index) {
+		this.index = index;
+	}
+
 	public List<Result> query(JSONObject jsonQuery) {
 		QueryBuilder qb = QueryBuilders.wrapperQuery(jsonQuery.toString());
 		LOG.debug(JsonUtils.prettify(jsonQuery));
@@ -30,8 +40,8 @@ public List<Result> query(JSONObject jsonQuery) {
 	}
 
 	private List<Result> query(QueryBuilder qb) {
-		SearchRequestBuilder searchRequestBuilder = client.prepareSearch().setQuery(qb).setSize(1000)
-				.addStoredField("pubmedId");
+		SearchRequestBuilder searchRequestBuilder = client.prepareSearch(index).setQuery(qb).setSize(1000)
+				.addStoredField("_id");
 
 		SearchResponse response = searchRequestBuilder.get();
 		LOG.trace(JsonUtils.prettify(response.toString()));

src/main/resources/templates/baseline-ct.json

@@ -0,0 +1,53 @@
+{
+	"bool": {
+		"must": [
+			{
+				"range": {
+					"minimum_age": {
+						"lte": {{age}}
+					}
+				}
+			},
+			{
+				"range": {
+					"maximum_age": {
+						"gte": {{age}}
+					}
+				}
+			},
+			{
+				"match": {
+					"sex": "{{sex}}"
+				}
+			}
+		],
+		"should": [
+			{
+				"multi_match": {
+					"query": "{{disease}}",
+					"fields": [
+						"title^2",
+						"summary",
+						"inclusion"
+					],
+					"tie_breaker": 0.3,
+					"type": "best_fields",
+					"boost": 1
+				}
+			},
+			{
+				"multi_match": {
+					"query": "{{gene}}",
+					"fields": [
+						"title^2",
+						"summary",
+						"inclusion"
+					],
+					"tie_breaker": 0.3,
+					"type": "best_fields",
+					"boost": 1
+				}
+			}
+		]
+	}
+}

src/test/java/at/medunigraz/imi/bst/trec/query/DummyElasticSearchQuery.java

@@ -7,11 +7,12 @@
 
 public class DummyElasticSearchQuery extends ElasticSearchQuery {
 
-	private Topic topic;
-
 	public DummyElasticSearchQuery() {
+		super("NOOP");
 	}
 
+	private Topic topic;
+
 	@Override
 	public List<Result> query(Topic topic) {
 		this.topic = topic;

src/test/java/at/medunigraz/imi/bst/trec/query/GeneExpanderQueryDecoratorTest.java

@@ -18,7 +18,7 @@ public class GeneExpanderQueryDecoratorTest extends QueryDecoratorTest {
 
 	public GeneExpanderQueryDecoratorTest() {
 		this.decoratedQuery = new GeneExpanderQueryDecorator(EXPAND_TO,
-				new TemplateQueryDecorator(template, new ElasticSearchQuery()));
+				new TemplateQueryDecorator(template, new ElasticSearchQuery("trec")));
 		this.topic = new Topic().withGene(GENE);
 	}
 

src/test/java/at/medunigraz/imi/bst/trec/query/TemplateQueryDecoratorTest.java

@@ -15,7 +15,7 @@ public class TemplateQueryDecoratorTest extends QueryDecoratorTest {
 	private final File template = new File(getClass().getResource("/templates/match-title.json").getFile());
 
 	public TemplateQueryDecoratorTest() {
-		this.decoratedQuery = new TemplateQueryDecorator(template, new ElasticSearchQuery());
+		this.decoratedQuery = new TemplateQueryDecorator(template, new ElasticSearchQuery("trec"));
 		this.topic = new Topic().withDisease(DISEASE);
 	}
 

src/test/java/at/medunigraz/imi/bst/trec/query/WordRemovalQueryDecoratorTest.java

@@ -16,7 +16,7 @@ public class WordRemovalQueryDecoratorTest extends QueryDecoratorTest {
 
 	public WordRemovalQueryDecoratorTest() {
 		this.decoratedQuery = new WordRemovalQueryDecorator(
-				new TemplateQueryDecorator(template, new ElasticSearchQuery()));
+				new TemplateQueryDecorator(template, new ElasticSearchQuery("trec")));
 		this.topic = new Topic().withDisease(DISEASE);
 	}