Examples with WeightedTerm org.ambraproject.rhino.service.taxonomy.WeightedTerm used on opensource projects

Example 1 with WeightedTerm

use of org.ambraproject.rhino.service.taxonomy.WeightedTerm in project rhino by PLOS.

the class TaxonomyClassificationServiceImplTest method testParseVectorElement.

@Test
public void testParseVectorElement() throws Exception {
    assertEquals(TaxonomyClassificationServiceImpl.parseVectorElement("<TERM>/Biology and life sciences/Computational biology/Computational neuroscience/Single neuron function|(5) neuron*(5)</TERM>"), new WeightedTerm("/Biology and life sciences/Computational biology/Computational neuroscience/Single neuron function", 5));
    assertEquals(TaxonomyClassificationServiceImpl.parseVectorElement("<TERM>/Medicine and health sciences/Anesthesiology/Anesthesia|(5) anesthesia(5)</TERM>"), new WeightedTerm("/Medicine and health sciences/Anesthesiology/Anesthesia", 5));
    assertEquals(TaxonomyClassificationServiceImpl.parseVectorElement("<TERM>/Medicine and health sciences/Geriatrics/Frailty|(19) frailty(18) frail*(1)</TERM>"), new WeightedTerm("/Medicine and health sciences/Geriatrics/Frailty", 19));
    assertEquals(TaxonomyClassificationServiceImpl.parseVectorElement("<TERM>/Biology and life sciences/Anatomy/Head/Face/Nose|(311) nose(311)</TERM>"), new WeightedTerm("/Biology and life sciences/Anatomy/Head/Face/Nose", 311));
    assertEquals(TaxonomyClassificationServiceImpl.parseVectorElement("<TERM>/People and places/Demography|(7) demographics(7)</TERM>"), new WeightedTerm("/People and places/Demography", 7));
    assertEquals(TaxonomyClassificationServiceImpl.parseVectorElement("<TERM>/Medicine and health sciences/Neurology/Cognitive neurology|(2) cognit*(2)</TERM>"), new WeightedTerm("/Medicine and health sciences/Neurology/Cognitive neurology", 2));
    assertEquals(TaxonomyClassificationServiceImpl.parseVectorElement("<TERM> /Medicine and health sciences/Neurology/Cognitive neurology| (67) cognit*(2)</TERM>"), new WeightedTerm("/Medicine and health sciences/Neurology/Cognitive neurology", 67));
}

Example 2 with WeightedTerm

use of org.ambraproject.rhino.service.taxonomy.WeightedTerm in project rhino by PLOS.

the class TaxonomyClassificationServiceImpl method persistCategories.

private void persistCategories(List<WeightedTerm> terms, Article article) {
    Set<String> termStrings = terms.stream().map(WeightedTerm::getPath).collect(Collectors.toSet());
    Collection<Category> existingCategories = hibernateTemplate.execute(session -> {
        Query query = session.createQuery("FROM Category WHERE path IN (:terms)");
        query.setParameterList("terms", termStrings);
        return (Collection<Category>) query.list();
    });
    Map<String, Category> existingCategoryMap = Maps.uniqueIndex(existingCategories, Category::getPath);
    Collection<ArticleCategoryAssignment> existingAssignments = getAssignmentsForArticle(article);
    Map<Category, ArticleCategoryAssignment> assignmentMap = Maps.uniqueIndex(existingAssignments, ArticleCategoryAssignment::getCategory);
    // Make it mutable. We will remove assignments as they are updated.
    assignmentMap = new HashMap<>(assignmentMap);
    for (WeightedTerm term : terms) {
        Category category = existingCategoryMap.get(term.getPath());
        if (category == null) {
            /*
         * A new category from the taxonomy server, which is not yet persisted in our system. Create it now.
         *
         * This risks a race condition if two articles are being populated concurrently and both have the same new
         * category, which can cause a "MySQLIntegrityConstraintViolationException: Duplicate entry" error.
         */
            category = new Category();
            category.setPath(term.getPath());
            hibernateTemplate.save(category);
        }
        ArticleCategoryAssignment assignment = assignmentMap.remove(category);
        if (assignment == null) {
            hibernateTemplate.save(new ArticleCategoryAssignment(category, article, term.getWeight()));
        } else {
            assignment.setWeight(term.getWeight());
            hibernateTemplate.update(assignment);
        }
    }
    // Each assignment that was not removed from assignmentMap is not among the new terms, so it should be deleted.
    assignmentMap.values().forEach(hibernateTemplate::delete);
}

Example 3 with WeightedTerm

use of org.ambraproject.rhino.service.taxonomy.WeightedTerm in project rhino by PLOS.

the class TaxonomyClassificationServiceImpl method classifyArticle.

/**
   * @inheritDoc
   */
@Override
public List<WeightedTerm> classifyArticle(Article article, Document articleXml) {
    RuntimeConfiguration.TaxonomyConfiguration configuration = getTaxonomyConfiguration();
    List<String> rawTerms = getRawTerms(articleXml, article, false);
    List<WeightedTerm> results = new ArrayList<>(rawTerms.size());
    for (String rawTerm : rawTerms) {
        WeightedTerm entry = parseVectorElement(rawTerm);
        String term = entry.getPath();
        if (term != null) {
            boolean isBlacklisted = false;
            for (String blacklistedCategory : configuration.getCategoryBlacklist()) {
                if (term.startsWith(blacklistedCategory)) {
                    isBlacklisted = true;
                    break;
                }
            }
            if (!isBlacklisted) {
                results.add(entry);
            }
        }
    }
    return results;
}

Example 4 with WeightedTerm

use of org.ambraproject.rhino.service.taxonomy.WeightedTerm in project rhino by PLOS.

the class TaxonomyClassificationServiceImpl method parseVectorElement.

/**
   * Parses a single line of the XML response from the taxonomy server.
   *
   * @param vectorElement The text body of a line of the response
   * @return the term and weight of the term
   */
@VisibleForTesting
static WeightedTerm parseVectorElement(String vectorElement) {
    Matcher match = TERM_PATTERN.matcher(vectorElement);
    if (match.find()) {
        String text = match.group(1);
        int value = Integer.parseInt(match.group(2));
        return new WeightedTerm(text, value);
    } else {
        //Bad term
        throw new TaxonomyRemoteServiceInvalidBehaviorException("Invalid syntax: " + vectorElement);
    }
}

Example 5 with WeightedTerm

use of org.ambraproject.rhino.service.taxonomy.WeightedTerm in project rhino by PLOS.

the class TaxonomyClassificationServiceImpl method populateCategories.

/**
   * {@inheritDoc}
   */
@Override
public void populateCategories(ArticleRevision revision) {
    ArticleIngestion ingestion = revision.getIngestion();
    Article article = ingestion.getArticle();
    Document xml = articleCrudService.getManuscriptXml(ingestion);
    List<WeightedTerm> terms;
    String doi = article.getDoi();
    //todo: fix or remove this when we find a home for article types
    boolean isAmendment = false;
    if (!isAmendment) {
        terms = classifyArticle(article, xml);
        if (terms != null && terms.size() > 0) {
            List<WeightedTerm> leafNodes = getDistinctLeafNodes(CATEGORY_COUNT, terms);
            persistCategories(leafNodes, article);
        } else {
            log.error("Taxonomy server returned 0 terms. Cannot populate Categories. " + doi);
        }
    }
}