Examples with Entity edu.stanford.muse.ner.Entity used on opensource projects

Example 1 with Entity

use of edu.stanford.muse.ner.Entity in project epadd by ePADD.

the class EntityBook method getDisplayNameToFreq.

public Map<String, Integer> getDisplayNameToFreq(Archive archive, short type) {
    Map<String, Entity> displayNameToEntity = new LinkedHashMap();
    double theta = 0.001;
    EntityBook entityBook = archive.getEntityBook();
    for (Document doc : archive.getAllDocs()) {
        Span[] spans = archive.getEntitiesInDoc(doc, true);
        Set<String> seenInThisDoc = new LinkedHashSet<>();
        for (Span span : spans) {
            // bail out if not of entity type that we're looking for, or not enough confidence
            if (span.type != type || span.typeScore < theta)
                continue;
            String name = span.getText();
            String displayName = name;
            // map the name to its display name. if no mapping, we should get the same name back as its displayName
            if (entityBook != null)
                displayName = entityBook.getDisplayName(name, span.type);
            displayName = displayName.trim();
            if (seenInThisDoc.contains(displayName))
                // count an entity in a doc only once
                continue;
            seenInThisDoc.add(displayName);
            if (!displayNameToEntity.containsKey(displayName))
                displayNameToEntity.put(displayName, new Entity(displayName, span.typeScore));
            else
                displayNameToEntity.get(displayName).freq++;
        }
    }
    // convert from displayNameToEntity to displayNameToFreq
    Map<String, Integer> displayNameToFreq = new LinkedHashMap<>();
    for (Entity e : displayNameToEntity.values()) displayNameToFreq.put(e.entity, e.freq);
    return displayNameToFreq;
}

Example 2 with Entity

use of edu.stanford.muse.ner.Entity in project epadd by ePADD.

the class EntityBook method fillSummaryFields.

public void fillSummaryFields(Map<MappedEntity, Pair<Double, Set<Document>>> docsetmap, Archive archive) {
    JSONArray resultArray = new JSONArray();
    // trick to use count (modifiable variable) inside for each.
    final Integer[] count = { 0 };
    summary_L1_entityCountMap.clear();
    docsetmap.entrySet().forEach(entry -> {
        count[0] = count[0] + 1;
        Summary_L1 summary = new Summary_L1();
        summary.score = entry.getValue().first;
        summary.messages = entry.getValue().second;
        // get date range
        Collection<EmailDocument> emaildocs = summary.messages.stream().map(s -> (EmailDocument) s).collect(Collectors.toList());
        Pair<Date, Date> daterange = EmailUtils.getFirstLast(emaildocs, true);
        if (daterange == null) {
            daterange = new Pair<>(archive.collectionMetadata.firstDate, archive.collectionMetadata.lastDate);
        }
        if (daterange.first == null)
            daterange.first = archive.collectionMetadata.firstDate;
        if (daterange.second == null)
            daterange.second = archive.collectionMetadata.lastDate;
        summary.startDate = daterange.first;
        summary.endDate = daterange.second;
        summary_L1_entityCountMap.put(entry.getKey(), summary);
        String entity = entry.getKey().getDisplayName();
        JSONArray j = new JSONArray();
        Short etype = entityType;
        Set<String> altNamesSet = entry.getKey().getAltNames();
        String altNames = (altNamesSet == null) ? "" : "Alternate names: " + Util.join(altNamesSet, ";");
        j.put(0, Util.escapeHTML(entity));
        j.put(1, summary.score);
        j.put(2, summary.messages.size());
        j.put(3, altNames);
        if (summary.startDate != null)
            j.put(4, new SimpleDateFormat("MM/dd/yyyy").format(summary.startDate));
        else
            j.put(4, summary.startDate);
        if (summary.endDate != null)
            j.put(5, new SimpleDateFormat("MM/dd/yyyy").format(summary.endDate));
        else
            j.put(5, summary.endDate);
        // add entity type as well..
        j.put(6, NEType.getTypeForCode(entityType).getDisplayName());
        resultArray.put(count[0] - 1, j);
    });
    summary_JSON = resultArray;
}