Examples of edu.stanford.nlp.util.CoreMap

edu.stanford.nlp.util.CoreMap

Base type for all annotatable core objects. Should usually be instantiated as {@link ArrayCoreMap}. Many common key definitions live in {@link edu.stanford.nlp.ling.CoreAnnotations}, but others may be defined elsewhere. See {@link edu.stanford.nlp.ling.CoreAnnotations} for details.

Note that implementations of this interface must take care to implement equality correctly: by default, two CoreMaps are .equal if they contain the same keys and all corresponding values are .equal. Subclasses that wish to change this behavior (such as {@link HashableCoreMap}) must make sure that all other CoreMap implementations have a special case in their .equals to use that equality definition when appropriate. Similarly, care must be taken when defining hashcodes. The default hashcode is 37 * sum of all keys' hashcodes plus the sum of all values' hashcodes. However, use of this class as HashMap keys is discouraged because the hashcode can change over time. Consider using a {@link HashableCoreMap}.
@author dramage @author rafferty

    assertFalse(match);


  }


  public void testTokenSequenceMatcher6() throws IOException {
    CoreMap doc = createDocument(testText1);
    TokenSequencePattern p = TokenSequencePattern.compile("[ /.*/ ] [ /.*/ ] [/of/] [/.*/]");
    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("first Bishop of London", m.group());
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("third Archbishop of Canterbury", m.group());
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("a member of the", m.group());
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("as Bishop of London", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile("([ /.*/ ] [ /.*/ ]) [/of/] ([/.*/])");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(2, m.groupCount());
    assertEquals("first Bishop of London", m.group());
    assertEquals("first Bishop", m.group(1));

View Full Code Here

    assertFalse(match);


  }


  public void testTokenSequenceMatcher7() throws IOException {
    CoreMap doc = createDocument(testText1);


    // Test sequence with groups
    TokenSequencePattern p = TokenSequencePattern.compile(" ( [ /[A-Za-z]+/ ]{1,2} )  [ /of/ ] ( [ /[A-Za-z]+/ ]{1,3} )");
    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(2, m.groupCount());
    assertEquals("first Bishop of London", m.group());
    assertEquals("first Bishop", m.group(1));
    assertEquals("London", m.group(2));
    match = m.find();
    assertTrue(match);
    assertEquals(2, m.groupCount());
    assertEquals("third Archbishop of Canterbury", m.group());
    assertEquals("third Archbishop", m.group(1));
    assertEquals("Canterbury", m.group(2));
    match = m.find();
    assertTrue(match);
    assertEquals(2, m.groupCount());
    assertEquals("a member of the Gregorian mission", m.group());
    assertEquals("a member", m.group(1));
    assertEquals("the Gregorian mission", m.group(2));
    match = m.find();
    assertTrue(match);
    assertEquals(2, m.groupCount());
    assertEquals("as Bishop of London in", m.group());
    assertEquals("as Bishop", m.group(1));
    assertEquals("London in", m.group(2));
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( " ( [ /[A-Za-z]+/ ]{2,2} )  [ /of/ ] ( [ /[A-Za-z]+/ ]{1,3}? )");


    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(2, m.groupCount());
    assertEquals("first Bishop of London", m.group());
    assertEquals("first Bishop", m.group(1));

View Full Code Here

    match = m.find();
    assertFalse(match);
  }


  public void testTokenSequenceMatcher8() throws IOException {
    CoreMap doc = createDocument(testText1);


    // Test sequence with groups
    TokenSequencePattern p = TokenSequencePattern.compile( "[ /[A-Za-z]+/ ]*");


    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("Mellitus was the first Bishop of London", m.group());
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("the third Archbishop of Canterbury", m.group());


    p = TokenSequencePattern.compile( "[ /[A-Za-z]+/ ]*  [\"Mellitus\"] [ \"was\"]");


    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("Mellitus was", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ /[A-Za-z]+/ ]+  [\"Mellitus\"] [ \"was\"]");


    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertFalse(match);


  }

View Full Code Here

    assertFalse(match);


  }


  public void testTokenSequenceMatcher9() throws IOException {
    CoreMap doc = createDocument(testText1);


    // Test sequence with groups
//    TokenSequencePattern p = TokenSequencePattern.compile( "(?$contextprev /.*/) (?$treat [{{treat}} & /.*/]) (?$contextnext [/.*/])");
    TokenSequencePattern p = TokenSequencePattern.compile("(?$contextprev /.*/) (?$test [{tag:NNP} & /.*/]) (?$contextnext [/.*/])");


    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(3, m.groupCount());
    assertEquals("first Bishop of", m.group());

View Full Code Here

    assertEquals(5, m.start("$contextnext"));
    assertEquals(6, m.end("$contextnext"));
  }


  public void testTokenSequenceMatcher10() throws IOException {
    CoreMap doc = createDocument("the number is five or 5 or 5.0 or but not 5x or -5 or 5L.");


    // Test simplified pattern with number
    TokenSequencePattern p = TokenSequencePattern.compile( "(five|5|5x|5.0|-5|5L)");


    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(1, m.groupCount());
    assertEquals("five", m.group(1));

View Full Code Here

    match = m.find();
    assertFalse(match);
  }


  public void testTokenSequenceOptimizeOrString() throws IOException {
    CoreMap doc = createDocument("atropine we need to have many many words here but we don't sweating");


    // Test simplified pattern with number
    TokenSequencePattern p = TokenSequencePattern.compile( "(?$dt \"atropine\") []{0,15} " +
            "(?$se  \"social\" \"avoidant\" \"behaviour\"|\"dysuria\"|\"hyperglycaemia\"| \"mental\" \"disorder\"|\"vertigo\"|\"flutter\"| \"chest\" \"pain\"| \"elevated\" \"blood\" \"pressure\"|\"mania\"| \"rash\" \"erythematous\"|\"manic\"| \"papular\" \"rash\"|\"death\"| \"atrial\" \"arrhythmia\"| \"dry\" \"eyes\"| \"loss\" \"of\" \"libido\"| \"rash\" \"papular\"|\"hypersensitivity\"| \"blood\" \"pressure\" \"increased\"|\"dyspepsia\"| \"accommodation\" \"disorder\"| \"reflexes\" \"increased\"|\"lesions\"|\"asthenia\"| \"gastrointestinal\" \"pain\"|\"excitement\"| \"breast\" \"feeding\"|\"hypokalaemia\"| \"cerebellar\" \"syndrome\"|\"nervousness\"| \"pulmonary\" \"oedema\"| \"inspiratory\" \"stridor\"| \"taste\" \"altered\"|\"paranoia\"| \"psychotic\" \"disorder\"| \"open\" \"angle\" \"glaucoma\"|\"photophobia\"| \"dry\" \"eye\"|\"osteoarthritis\"| \"keratoconjunctivitis\" \"sicca\"| \"haemoglobin\" \"increased\"| \"ventricular\" \"extrasystoles\"|\"hallucinations\"|\"conjunctivitis\"|\"paralysis\"| \"qrs\" \"complex\"|\"anxiety\"| \"conjunctival\" \"disorder\"|\"coma\"|\"strabismus\"|\"thirst\"|\"para\"| \"sicca\" \"syndrome\"| \"atrioventricular\" \"dissociation\"|\"desquamation\"|\"crusting\"| \"abdominal\" \"distension\"|\"blindness\"|\"hypotension\"|\"dermatitis\"| \"sinus\" \"tachycardia\"| \"abdominal\" \"distention\"| \"lacrimation\" \"decreased\"|\"sicca\"| \"paralytic\" \"ileus\"| \"urinary\" \"hesitation\"|\"withdrawn\"| \"erectile\" \"dysfunction\"|\"keratoconjunctivitis\"|\"anaphylaxis\"| \"psychiatric\" \"disorders\"| \"altered\" \"taste\"|\"somnolence\"|\"extrasystoles\"|\"ageusia\"| \"intraocular\" \"pressure\" \"increased\"| \"left\" \"ventricular\" \"failure\"|\"impotence\"|\"drowsiness\"|\"conjunctiva\"| \"delayed\" \"gastric\" \"emptying\"| \"gastrointestinal\" \"sounds\" \"abnormal\"| \"qt\" \"prolonged\"| \"supraventricular\" \"tachycardia\"|\"weakness\"|\"hypertonia\"| \"confusional\" \"state\"|\"anhidrosis\"|\"myopia\"|\"dyspnoea\"| \"speech\" \"impairment\" \"nos\"| \"rash\" \"maculo\" \"papular\"|\"petechiae\"|\"tachypnea\"| \"acute\" \"angle\" \"closure\" \"glaucoma\"| \"gastrooesophageal\" \"reflux\" \"disease\"|\"hypokalemia\"| \"left\" \"heart\" \"failure\"| \"myocardial\" \"infarction\"| \"site\" \"reaction\"| \"ventricular\" \"fibrillation\"|\"fibrillation\"| \"maculopapular\" \"rash\"| \"impaired\" \"gastric\" \"emptying\"|\"amnesia\"| \"labored\" \"respirations\"| \"decreased\" \"lacrimation\"|\"mydriasis\"|\"headache\"| \"dry\" \"mouth\"|\"scab\"| \"cardiac\" \"syncope\"| \"visual\" \"acuity\" \"reduced\"|\"tension\"| \"blurred\" \"vision\"| \"bloated\" \"feeling\"| \"labored\" \"breathing\"| \"stridor\" \"inspiratory\"| \"skin\" \"exfoliation\"| \"memory\" \"loss\"|\"syncope\"| \"rash\" \"scarlatiniform\"|\"hyperpyrexia\"| \"cardiac\" \"flutter\"|\"heartburn\"| \"bowel\" \"sounds\" \"decreased\"|\"blepharitis\"|\"tachycardia\"| \"excessive\" \"thirst\"|\"confusion\"| \"rash\" \"macular\"| \"taste\" \"loss\"| \"respiratory\" \"failure\"|\"hesitancy\"|\"dysmetria\"|\"disorientation\"| \"decreased\" \"hemoglobin\"| \"atrial\" \"fibrillation\"| \"urinary\" \"retention\"| \"dry\" \"skin\"|\"dehydration\"|\"hyponatraemia\"|\"dysgeusia\"|\"disorder\"| \"increased\" \"intraocular\" \"pressure\"| \"speech\" \"disorder\"| \"feeling\" \"abnormal\"|\"pain\"| \"anaphylactic\" \"shock\"|\"hallucination\"| \"abdominal\" \"pain\"| \"junctional\" \"tachycardia\"| \"bun\" \"increased\"| \"ventricular\" \"flutter\"| \"scarlatiniform\" \"rash\"|\"agitation\"| \"feeling\" \"hot\"|\"hyponatremia\"| \"decreased\" \"bowel\" \"sounds\"|\"cyanosis\"|\"dysarthria\"| \"heat\" \"intolerance\"|\"hyperglycemia\"|\"reflux\"| \"angle\" \"closure\" \"glaucoma\"| \"electrocardiogram\" \"qt\" \"prolonged\"| \"vision\" \"blurred\"| \"blood\" \"urea\" \"increased\"|\"dizziness\"|\"arrhythmia\"|\"erythema\"|\"vomiting\"| \"difficulty\" \"in\" \"micturition\"|\"infarction\"|\"laryngospasm\"|\"hypoglycaemia\"|\"hypoglycemia\"| \"elevated\" \"hemoglobin\"| \"skin\" \"warm\"| \"ventricular\" \"arrhythmia\"|\"dissociation\"| \"warm\" \"skin\"| \"follicular\" \"conjunctivitis\"|\"urticaria\"|\"fatigue\"| \"cardiac\" \"fibrillation\"| \"decreased\" \"sweating\"| \"decreased\" \"visual\" \"acuity\"|\"lethargy\"| \"acute\" \"angle\" \"closure\" \"glaucoma\"| \"nodal\" \"rhythm\"|\"borborygmi\"|\"hyperreflexia\"| \"respiratory\" \"depression\"|\"diarrhea\"|\"leukocytosis\"| \"speech\" \"disturbance\"|\"ataxia\"|\"cycloplegia\"|\"tachypnoea\"|\"eczema\"| \"supraventricular\" \"extrasystoles\"|\"ileus\"| \"cardiac\" \"arrest\"| \"ventricular\" \"tachycardia\"|\"laryngitis\"|\"delirium\"|\"lactation\"|\"glaucoma\"|\"obstruction\"|\"hypohidrosis\"|\"parity\"|\"palpitations\"| \"temperature\" \"intolerance\"|\"constipation\"|\"cyclophoria\"| \"acute\" \"coronary\" \"syndrome\"| \"arrhythmia\" \"supraventricular\"|\"arrest\"|\"lesion\"|\"nausea\"| \"sweating\" \"decreased\"|\"keratitis\"|\"dyskinesia\"| \"pulmonary\" \"function\" \"test\" \"decreased\"|\"stridor\"|\"swelling\"|\"dysphagia\"| \"haemoglobin\" \"decreased\"|\"diarrhoea\"| \"ileus\" \"paralytic\"|\"clonus\"|\"insomnia\"| \"electrocardiogram\" \"qrs\" \"complex\"| \"nasal\" \"congestion\"| \"nasal\" \"dryness\"|\"sweating\"|\"rash\"| \"nodal\" \"arrhythmia\"|\"irritability\"|\"hyperhidrosis\"| \"ventricular\" \"failure\")");


    Timing timing = new Timing();
    timing.start();
    for (int i = 0; i < 100; i++) {
      TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
      boolean match = m.find();
      assertTrue(match);
      assertEquals("atropine we need to have many many words here but we do n't sweating", m.group(0));


      match = m.find();
      assertFalse(match);
    }
    timing.stop("testTokenSequenceOptimizeOrString matched");




    CoreMap docNoMatch = createDocument("atropine we need to have many many words here but we don't, many many many words but still no match");
    timing.start();
    for (int i = 0; i < 100; i++) {
      TokenSequenceMatcher m = p.getMatcher(docNoMatch.get(CoreAnnotations.TokensAnnotation.class));
      boolean match = m.find();
      assertFalse(match);
    }
    timing.stop("testTokenSequenceOptimizeOrString no match");
  }

View Full Code Here

  public void testMultiplePatterns() throws IOException {
    TokenSequencePattern p1 = TokenSequencePattern.compile("(?$dt \"atropine\") []{0,15} " +
        "(?$se  \"social\" \"avoidant\" \"behaviour\"|\"dysuria\"|\"hyperglycaemia\"| \"mental\" \"disorder\"|\"vertigo\"|\"flutter\"| \"chest\" \"pain\"| \"elevated\" \"blood\" \"pressure\"|\"mania\"| \"rash\" \"erythematous\"|\"manic\"| \"papular\" \"rash\"|\"death\"| \"atrial\" \"arrhythmia\"| \"dry\" \"eyes\"| \"loss\" \"of\" \"libido\"| \"rash\" \"papular\"|\"hypersensitivity\"| \"blood\" \"pressure\" \"increased\"|\"dyspepsia\"| \"accommodation\" \"disorder\"| \"reflexes\" \"increased\"|\"lesions\"|\"asthenia\"| \"gastrointestinal\" \"pain\"|\"excitement\"| \"breast\" \"feeding\"|\"hypokalaemia\"| \"cerebellar\" \"syndrome\"|\"nervousness\"| \"pulmonary\" \"oedema\"| \"inspiratory\" \"stridor\"| \"taste\" \"altered\"|\"paranoia\"| \"psychotic\" \"disorder\"| \"open\" \"angle\" \"glaucoma\"|\"photophobia\"| \"dry\" \"eye\"|\"osteoarthritis\"| \"keratoconjunctivitis\" \"sicca\"| \"haemoglobin\" \"increased\"| \"ventricular\" \"extrasystoles\"|\"hallucinations\"|\"conjunctivitis\"|\"paralysis\"| \"qrs\" \"complex\"|\"anxiety\"| \"conjunctival\" \"disorder\"|\"coma\"|\"strabismus\"|\"thirst\"|\"para\"| \"sicca\" \"syndrome\"| \"atrioventricular\" \"dissociation\"|\"desquamation\"|\"crusting\"| \"abdominal\" \"distension\"|\"blindness\"|\"hypotension\"|\"dermatitis\"| \"sinus\" \"tachycardia\"| \"abdominal\" \"distention\"| \"lacrimation\" \"decreased\"|\"sicca\"| \"paralytic\" \"ileus\"| \"urinary\" \"hesitation\"|\"withdrawn\"| \"erectile\" \"dysfunction\"|\"keratoconjunctivitis\"|\"anaphylaxis\"| \"psychiatric\" \"disorders\"| \"altered\" \"taste\"|\"somnolence\"|\"extrasystoles\"|\"ageusia\"| \"intraocular\" \"pressure\" \"increased\"| \"left\" \"ventricular\" \"failure\"|\"impotence\"|\"drowsiness\"|\"conjunctiva\"| \"delayed\" \"gastric\" \"emptying\"| \"gastrointestinal\" \"sounds\" \"abnormal\"| \"qt\" \"prolonged\"| \"supraventricular\" \"tachycardia\"|\"weakness\"|\"hypertonia\"| \"confusional\" \"state\"|\"anhidrosis\"|\"myopia\"|\"dyspnoea\"| \"speech\" \"impairment\" \"nos\"| \"rash\" \"maculo\" \"papular\"|\"petechiae\"|\"tachypnea\"| \"acute\" \"angle\" \"closure\" \"glaucoma\"| \"gastrooesophageal\" \"reflux\" \"disease\"|\"hypokalemia\"| \"left\" \"heart\" \"failure\"| \"myocardial\" \"infarction\"| \"site\" \"reaction\"| \"ventricular\" \"fibrillation\"|\"fibrillation\"| \"maculopapular\" \"rash\"| \"impaired\" \"gastric\" \"emptying\"|\"amnesia\"| \"labored\" \"respirations\"| \"decreased\" \"lacrimation\"|\"mydriasis\"|\"headache\"| \"dry\" \"mouth\"|\"scab\"| \"cardiac\" \"syncope\"| \"visual\" \"acuity\" \"reduced\"|\"tension\"| \"blurred\" \"vision\"| \"bloated\" \"feeling\"| \"labored\" \"breathing\"| \"stridor\" \"inspiratory\"| \"skin\" \"exfoliation\"| \"memory\" \"loss\"|\"syncope\"| \"rash\" \"scarlatiniform\"|\"hyperpyrexia\"| \"cardiac\" \"flutter\"|\"heartburn\"| \"bowel\" \"sounds\" \"decreased\"|\"blepharitis\"|\"tachycardia\"| \"excessive\" \"thirst\"|\"confusion\"| \"rash\" \"macular\"| \"taste\" \"loss\"| \"respiratory\" \"failure\"|\"hesitancy\"|\"dysmetria\"|\"disorientation\"| \"decreased\" \"hemoglobin\"| \"atrial\" \"fibrillation\"| \"urinary\" \"retention\"| \"dry\" \"skin\"|\"dehydration\"|\"hyponatraemia\"|\"dysgeusia\"|\"disorder\"| \"increased\" \"intraocular\" \"pressure\"| \"speech\" \"disorder\"| \"feeling\" \"abnormal\"|\"pain\"| \"anaphylactic\" \"shock\"|\"hallucination\"| \"abdominal\" \"pain\"| \"junctional\" \"tachycardia\"| \"bun\" \"increased\"| \"ventricular\" \"flutter\"| \"scarlatiniform\" \"rash\"|\"agitation\"| \"feeling\" \"hot\"|\"hyponatremia\"| \"decreased\" \"bowel\" \"sounds\"|\"cyanosis\"|\"dysarthria\"| \"heat\" \"intolerance\"|\"hyperglycemia\"|\"reflux\"| \"angle\" \"closure\" \"glaucoma\"| \"electrocardiogram\" \"qt\" \"prolonged\"| \"vision\" \"blurred\"| \"blood\" \"urea\" \"increased\"|\"dizziness\"|\"arrhythmia\"|\"erythema\"|\"vomiting\"| \"difficulty\" \"in\" \"micturition\"|\"infarction\"|\"laryngospasm\"|\"hypoglycaemia\"|\"hypoglycemia\"| \"elevated\" \"hemoglobin\"| \"skin\" \"warm\"| \"ventricular\" \"arrhythmia\"|\"dissociation\"| \"warm\" \"skin\"| \"follicular\" \"conjunctivitis\"|\"urticaria\"|\"fatigue\"| \"cardiac\" \"fibrillation\"| \"decreased\" \"sweating\"| \"decreased\" \"visual\" \"acuity\"|\"lethargy\"| \"acute\" \"angle\" \"closure\" \"glaucoma\"| \"nodal\" \"rhythm\"|\"borborygmi\"|\"hyperreflexia\"| \"respiratory\" \"depression\"|\"diarrhea\"|\"leukocytosis\"| \"speech\" \"disturbance\"|\"ataxia\"|\"cycloplegia\"|\"tachypnoea\"|\"eczema\"| \"supraventricular\" \"extrasystoles\"|\"ileus\"| \"cardiac\" \"arrest\"| \"ventricular\" \"tachycardia\"|\"laryngitis\"|\"delirium\"|\"lactation\"|\"glaucoma\"|\"obstruction\"|\"hypohidrosis\"|\"parity\"|\"palpitations\"| \"temperature\" \"intolerance\"|\"constipation\"|\"cyclophoria\"| \"acute\" \"coronary\" \"syndrome\"| \"arrhythmia\" \"supraventricular\"|\"arrest\"|\"lesion\"|\"nausea\"| \"sweating\" \"decreased\"|\"keratitis\"|\"dyskinesia\"| \"pulmonary\" \"function\" \"test\" \"decreased\"|\"stridor\"|\"swelling\"|\"dysphagia\"| \"haemoglobin\" \"decreased\"|\"diarrhoea\"| \"ileus\" \"paralytic\"|\"clonus\"|\"insomnia\"| \"electrocardiogram\" \"qrs\" \"complex\"| \"nasal\" \"congestion\"| \"nasal\" \"dryness\"|\"sweating\"|\"rash\"| \"nodal\" \"arrhythmia\"|\"irritability\"|\"hyperhidrosis\"| \"ventricular\" \"failure\")");
    TokenSequencePattern p2 = TokenSequencePattern.compile( "(?$dt \"disease\") []{0,15} " +
            "(?$se  \"social\" \"avoidant\" \"behaviour\"|\"dysuria\"|\"hyperglycaemia\"| \"mental\" \"disorder\"|\"vertigo\"|\"flutter\"| \"chest\" \"pain\"| \"elevated\" \"blood\" \"pressure\"|\"mania\"| \"rash\" \"erythematous\"|\"manic\"| \"papular\" \"rash\"|\"death\"| \"atrial\" \"arrhythmia\"| \"dry\" \"eyes\"| \"loss\" \"of\" \"libido\"| \"rash\" \"papular\"|\"hypersensitivity\"| \"blood\" \"pressure\" \"increased\"|\"dyspepsia\"| \"accommodation\" \"disorder\"| \"reflexes\" \"increased\"|\"lesions\"|\"asthenia\"| \"gastrointestinal\" \"pain\"|\"excitement\"| \"breast\" \"feeding\"|\"hypokalaemia\"| \"cerebellar\" \"syndrome\"|\"nervousness\"| \"pulmonary\" \"oedema\"| \"inspiratory\" \"stridor\"| \"taste\" \"altered\"|\"paranoia\"| \"psychotic\" \"disorder\"| \"open\" \"angle\" \"glaucoma\"|\"photophobia\"| \"dry\" \"eye\"|\"osteoarthritis\"| \"keratoconjunctivitis\" \"sicca\"| \"haemoglobin\" \"increased\"| \"ventricular\" \"extrasystoles\"|\"hallucinations\"|\"conjunctivitis\"|\"paralysis\"| \"qrs\" \"complex\"|\"anxiety\"| \"conjunctival\" \"disorder\"|\"coma\"|\"strabismus\"|\"thirst\"|\"para\"| \"sicca\" \"syndrome\"| \"atrioventricular\" \"dissociation\"|\"desquamation\"|\"crusting\"| \"abdominal\" \"distension\"|\"blindness\"|\"hypotension\"|\"dermatitis\"| \"sinus\" \"tachycardia\"| \"abdominal\" \"distention\"| \"lacrimation\" \"decreased\"|\"sicca\"| \"paralytic\" \"ileus\"| \"urinary\" \"hesitation\"|\"withdrawn\"| \"erectile\" \"dysfunction\"|\"keratoconjunctivitis\"|\"anaphylaxis\"| \"psychiatric\" \"disorders\"| \"altered\" \"taste\"|\"somnolence\"|\"extrasystoles\"|\"ageusia\"| \"intraocular\" \"pressure\" \"increased\"| \"left\" \"ventricular\" \"failure\"|\"impotence\"|\"drowsiness\"|\"conjunctiva\"| \"delayed\" \"gastric\" \"emptying\"| \"gastrointestinal\" \"sounds\" \"abnormal\"| \"qt\" \"prolonged\"| \"supraventricular\" \"tachycardia\"|\"weakness\"|\"hypertonia\"| \"confusional\" \"state\"|\"anhidrosis\"|\"myopia\"|\"dyspnoea\"| \"speech\" \"impairment\" \"nos\"| \"rash\" \"maculo\" \"papular\"|\"petechiae\"|\"tachypnea\"| \"acute\" \"angle\" \"closure\" \"glaucoma\"| \"gastrooesophageal\" \"reflux\" \"disease\"|\"hypokalemia\"| \"left\" \"heart\" \"failure\"| \"myocardial\" \"infarction\"| \"site\" \"reaction\"| \"ventricular\" \"fibrillation\"|\"fibrillation\"| \"maculopapular\" \"rash\"| \"impaired\" \"gastric\" \"emptying\"|\"amnesia\"| \"labored\" \"respirations\"| \"decreased\" \"lacrimation\"|\"mydriasis\"|\"headache\"| \"dry\" \"mouth\"|\"scab\"| \"cardiac\" \"syncope\"| \"visual\" \"acuity\" \"reduced\"|\"tension\"| \"blurred\" \"vision\"| \"bloated\" \"feeling\"| \"labored\" \"breathing\"| \"stridor\" \"inspiratory\"| \"skin\" \"exfoliation\"| \"memory\" \"loss\"|\"syncope\"| \"rash\" \"scarlatiniform\"|\"hyperpyrexia\"| \"cardiac\" \"flutter\"|\"heartburn\"| \"bowel\" \"sounds\" \"decreased\"|\"blepharitis\"|\"tachycardia\"| \"excessive\" \"thirst\"|\"confusion\"| \"rash\" \"macular\"| \"taste\" \"loss\"| \"respiratory\" \"failure\"|\"hesitancy\"|\"dysmetria\"|\"disorientation\"| \"decreased\" \"hemoglobin\"| \"atrial\" \"fibrillation\"| \"urinary\" \"retention\"| \"dry\" \"skin\"|\"dehydration\"|\"hyponatraemia\"|\"dysgeusia\"|\"disorder\"| \"increased\" \"intraocular\" \"pressure\"| \"speech\" \"disorder\"| \"feeling\" \"abnormal\"|\"pain\"| \"anaphylactic\" \"shock\"|\"hallucination\"| \"abdominal\" \"pain\"| \"junctional\" \"tachycardia\"| \"bun\" \"increased\"| \"ventricular\" \"flutter\"| \"scarlatiniform\" \"rash\"|\"agitation\"| \"feeling\" \"hot\"|\"hyponatremia\"| \"decreased\" \"bowel\" \"sounds\"|\"cyanosis\"|\"dysarthria\"| \"heat\" \"intolerance\"|\"hyperglycemia\"|\"reflux\"| \"angle\" \"closure\" \"glaucoma\"| \"electrocardiogram\" \"qt\" \"prolonged\"| \"vision\" \"blurred\"| \"blood\" \"urea\" \"increased\"|\"dizziness\"|\"arrhythmia\"|\"erythema\"|\"vomiting\"| \"difficulty\" \"in\" \"micturition\"|\"infarction\"|\"laryngospasm\"|\"hypoglycaemia\"|\"hypoglycemia\"| \"elevated\" \"hemoglobin\"| \"skin\" \"warm\"| \"ventricular\" \"arrhythmia\"|\"dissociation\"| \"warm\" \"skin\"| \"follicular\" \"conjunctivitis\"|\"urticaria\"|\"fatigue\"| \"cardiac\" \"fibrillation\"| \"decreased\" \"sweating\"| \"decreased\" \"visual\" \"acuity\"|\"lethargy\"| \"acute\" \"angle\" \"closure\" \"glaucoma\"| \"nodal\" \"rhythm\"|\"borborygmi\"|\"hyperreflexia\"| \"respiratory\" \"depression\"|\"diarrhea\"|\"leukocytosis\"| \"speech\" \"disturbance\"|\"ataxia\"|\"cycloplegia\"|\"tachypnoea\"|\"eczema\"| \"supraventricular\" \"extrasystoles\"|\"ileus\"| \"cardiac\" \"arrest\"| \"ventricular\" \"tachycardia\"|\"laryngitis\"|\"delirium\"|\"lactation\"|\"glaucoma\"|\"obstruction\"|\"hypohidrosis\"|\"parity\"|\"palpitations\"| \"temperature\" \"intolerance\"|\"constipation\"|\"cyclophoria\"| \"acute\" \"coronary\" \"syndrome\"| \"arrhythmia\" \"supraventricular\"|\"arrest\"|\"lesion\"|\"nausea\"| \"sweating\" \"decreased\"|\"keratitis\"|\"dyskinesia\"| \"pulmonary\" \"function\" \"test\" \"decreased\"|\"stridor\"|\"swelling\"|\"dysphagia\"| \"haemoglobin\" \"decreased\"|\"diarrhoea\"| \"ileus\" \"paralytic\"|\"clonus\"|\"insomnia\"| \"electrocardiogram\" \"qrs\" \"complex\"| \"nasal\" \"congestion\"| \"nasal\" \"dryness\"|\"sweating\"|\"rash\"| \"nodal\" \"arrhythmia\"|\"irritability\"|\"hyperhidrosis\"| \"ventricular\" \"failure\")");
    CoreMap doc = createDocument("atropine we need to have many many words here but we don't sweating");
    MultiPatternMatcher<CoreMap> multiPatternMatcher = TokenSequencePattern.getMultiPatternMatcher(p1, p2);
    List<String> expected = new ArrayList<String>();
    expected.add("atropine we need to have many many words here but we do n't sweating");
    Iterator<String> expectedIter = expected.iterator();


    Iterable<SequenceMatchResult<CoreMap>> matches =
            multiPatternMatcher.findAllNonOverlappingMatchesPerPattern(doc.get(CoreAnnotations.TokensAnnotation.class));
    for (SequenceMatchResult<CoreMap> match:matches) {
     assertEquals(expectedIter.next(), match.group());
    }
    assertFalse(expectedIter.hasNext());
  }

View Full Code Here

    }
    assertFalse(expectedIter.hasNext());
  }


  public void testTokenSequenceMatcherPosNNP() throws IOException {
    CoreMap doc = createDocument(testText1);


    // Test sequence with groups
    TokenSequencePattern p = TokenSequencePattern.compile( "[ { tag:\"NNP\" } ]+");
    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("Mellitus", m.group());


    p = TokenSequencePattern.compile( "[ { tag:\"NNP\" } ] [ /is|was/ ] []*? [ { tag:\"NNP\" } ]+ ");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("Mellitus was the first Bishop", m.group());


    TokenSequencePattern nnpPattern = TokenSequencePattern.compile( "[ { tag:\"NNP\" } ]" );
    Env env = TokenSequencePattern.getNewEnv();
    env.bind("$NNP", nnpPattern);
    p = TokenSequencePattern.compile(env, " $NNP [ /is|was/ ] []*? $NNP+ [ \"of\" ] $NNP+ ");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("Mellitus was the first Bishop of London", m.group());


    p = TokenSequencePattern.compile(env, " ($NNP) /is|was/ []*? ($NNP)+ \"of\" ($NNP)+ ");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(3, m.groupCount());
    assertEquals("Mellitus was the first Bishop of London", m.group());
    assertEquals("Mellitus", m.group(1));
    assertEquals("Bishop", m.group(2));
    assertEquals("London", m.group(3));




    nnpPattern = TokenSequencePattern.compile( " ( [ { tag:\"NNP\" } ] )" );
    env.bind("$NNP", nnpPattern);
    p = TokenSequencePattern.compile(env, " $NNP /is|was/ []*? $NNP+ \"of\" $NNP+ ");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(3, m.groupCount());
    assertEquals("Mellitus was the first Bishop of London", m.group());
    assertEquals("Mellitus", m.group(1));
    assertEquals("Bishop", m.group(2));
    assertEquals("London", m.group(3));




    // Same as above but without extra "{}"
    nnpPattern = TokenSequencePattern.compile( " ( [ tag:\"NNP\" ] )" );
    env.bind("$NNP", nnpPattern);
    p = TokenSequencePattern.compile(env, " $NNP /is|was/ []*? $NNP+ \"of\" $NNP+ ");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(3, m.groupCount());
    assertEquals("Mellitus was the first Bishop of London", m.group());
    assertEquals("Mellitus", m.group(1));
    assertEquals("Bishop", m.group(2));
    assertEquals("London", m.group(3));


    // Same as above but using "pos"
    nnpPattern = TokenSequencePattern.compile( " ( [ pos:\"NNP\" ] )" );
    env.bind("$NNP", nnpPattern);
    p = TokenSequencePattern.compile(env, " $NNP /is|was/ []*? $NNP+ \"of\" $NNP+ ");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(3, m.groupCount());
    assertEquals("Mellitus was the first Bishop of London", m.group());
    assertEquals("Mellitus", m.group(1));

View Full Code Here

    assertEquals("Bishop", m.group(2));
    assertEquals("London", m.group(3));
  }


  public void testTokenSequenceMatcherNumber() throws IOException {
    CoreMap doc = createDocument("It happened on January 3, 2002");


    // Test sequence with groups
    TokenSequencePattern p = TokenSequencePattern.compile( "[ { word::IS_NUM } ]+");
    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("3", m.group());
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("2002", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { word>=2002 } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("2002", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { word>2002 } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertFalse(match);


    // Check no {} with or
    p = TokenSequencePattern.compile( "[ word > 2002 | word==2002 ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("2002", m.group());
    match = m.find();
    assertFalse(match);


    // Check no {} with and
    p = TokenSequencePattern.compile( "[ word>2002 & word==2002 ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { word>2000 } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("2002", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { word<=2002 } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("3", m.group());
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("2002", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { word<2002 } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("3", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { word==2002 } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("2002", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { ner:DATE } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("January 3 , 2002", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { ner::NOT_NIL } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("January 3 , 2002", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ { ner::IS_NIL } ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("It happened on", m.group());
    match = m.find();
    assertFalse(match);


    p = TokenSequencePattern.compile( "[ {{ word=~/2002/ }} ]+");
    m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    match = m.find();
    assertTrue(match);
    assertEquals(0, m.groupCount());
    assertEquals("2002", m.group());
    match = m.find();

View Full Code Here

    match = m.find();
    assertFalse(match);
  }


  public void testTokenSequenceMatcherNested() throws IOException {
    CoreMap doc = createDocument("A A A B B B B B B C C");


    // Test sequence with groups
    TokenSequencePattern p = TokenSequencePattern.compile( "( /B/+ )+");
    TokenSequenceMatcher m = p.getMatcher(doc.get(CoreAnnotations.TokensAnnotation.class));
    boolean match = m.find();
    assertTrue(match);
    assertEquals(1, m.groupCount());
    assertEquals("B B B B B B", m.group());
    assertEquals("B B B B B B", m.group(1));

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of edu.stanford.nlp.util.CoreMap

edu.stanford.nlp.dcoref.ACEMentionExtractor

edu.stanford.nlp.dcoref.CoNLLMentionExtractor

edu.stanford.nlp.dcoref.Document

edu.stanford.nlp.dcoref.MUCMentionExtractor

edu.stanford.nlp.dcoref.RuleBasedCorefMentionFinder

edu.stanford.nlp.dcoref.SieveCoreferenceSystem

edu.stanford.nlp.ie.crf.TestSequenceModel

edu.stanford.nlp.ie.machinereading.BasicRelationFeatureFactory

edu.stanford.nlp.ie.machinereading.domains.ace.AceReader

edu.stanford.nlp.ie.machinereading.EntityExtractorResultsPrinter

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.