|
0
|
1 package de.mpiwg.anteater.persons.regex;
|
|
|
2
|
|
|
3 import java.util.ArrayList;
|
|
|
4 import java.util.HashMap;
|
|
|
5 import java.util.List;
|
|
|
6 import java.util.Map;
|
|
|
7 import java.util.regex.Matcher;
|
|
|
8 import java.util.regex.Pattern;
|
|
|
9
|
|
|
10 import de.mpiwg.anteater.AnteaterConfiguration;
|
|
|
11 import de.mpiwg.anteater.persons.APerson;
|
|
|
12 import de.mpiwg.anteater.persons.Person;
|
|
|
13 import de.mpiwg.anteater.persons.PersonsExtraction;
|
|
|
14 import de.mpiwg.anteater.results.ApplicantResult;
|
|
|
15 import de.mpiwg.anteater.text.Paragraph;
|
|
|
16 import de.mpiwg.anteater.text.TextInformation;
|
|
|
17 import de.mpiwg.anteater.text.TextPart;
|
|
|
18 import de.mpiwg.anteater.text.TextType;
|
|
|
19
|
|
|
20 public class ApplicantRegexFinder {
|
|
|
21
|
|
|
22 public final static String COMPONENT_NAME = ApplicantRegexFinder.class.getSimpleName();
|
|
|
23
|
|
|
24 private String applicantRegex = "Applicant:\\p{Blank}{0,1}([A-Z]{2}-.+?,){0,1}\\p{Blank}{0,1}(.+?,( {0,1}Inc(.){0,1}| {0,1}LLC){0,1})";
|
|
|
25
|
|
|
26 private AnteaterConfiguration configuration;
|
|
|
27
|
|
|
28 public ApplicantRegexFinder(AnteaterConfiguration configuration) {
|
|
|
29 this.configuration = configuration;
|
|
|
30 }
|
|
|
31
|
|
|
32 public List<ApplicantResult> findApplicants(List<TextInformation> infos) {
|
|
|
33 configuration.getLogger().logMessage(COMPONENT_NAME, "Search for applicants with regular expressions.");
|
|
|
34
|
|
|
35 List<ApplicantResult> results = new ArrayList<ApplicantResult>();
|
|
|
36
|
|
|
37 for (TextInformation info : infos) {
|
|
|
38 List<PersonsExtraction> extractions = info.getPersonsExtractions();
|
|
|
39
|
|
|
40
|
|
|
41 // find applicants in summaries
|
|
|
42 List<TextPart> summaries = info.getSummaries();
|
|
|
43 for (TextPart summary : summaries) {
|
|
|
44
|
|
|
45 PersonsExtraction pExtraction =findExtraction(extractions, TextType.TYPE_SUMMARY, summaries.indexOf(summary));
|
|
|
46 results.addAll(findApplicantsInText(info, pExtraction, summary));
|
|
|
47 }
|
|
|
48
|
|
|
49 // find applicants in suppleInf
|
|
|
50 List<TextPart> suppleInf = info.getSupplInfos();
|
|
|
51 for (TextPart sInf : suppleInf) {
|
|
|
52 PersonsExtraction pExtraction = findExtraction(extractions, TextType.TYPE_SUPLINF, suppleInf.indexOf(sInf));
|
|
|
53 results.addAll(findApplicantsInText(info, pExtraction, sInf));
|
|
|
54 }
|
|
|
55 }
|
|
|
56
|
|
|
57 configuration.getLogger().logMessage(COMPONENT_NAME, "Found " + results.size() + " applicant(s).");
|
|
|
58
|
|
|
59 return results;
|
|
|
60 }
|
|
|
61
|
|
|
62 private PersonsExtraction findExtraction(List<PersonsExtraction> extractions, int textType, int textIdx) {
|
|
|
63 PersonsExtraction pExtraction = null;
|
|
|
64
|
|
|
65 for (PersonsExtraction extr : extractions) {
|
|
|
66 if (extr.getType() == textType && extr.getTextIdx() == textIdx) {
|
|
|
67 pExtraction = extr;
|
|
|
68 break;
|
|
|
69 }
|
|
|
70 }
|
|
|
71
|
|
|
72 if (pExtraction == null) {
|
|
|
73 pExtraction = new PersonsExtraction();
|
|
|
74 pExtraction.setType(textType);
|
|
|
75 pExtraction.setTextIdx(textIdx);
|
|
|
76 pExtraction.setPerson(new ArrayList<APerson>());
|
|
|
77 extractions.add(pExtraction);
|
|
|
78 }
|
|
|
79
|
|
|
80 return pExtraction;
|
|
|
81 }
|
|
|
82
|
|
|
83 private List<ApplicantResult> findApplicantsInText(TextInformation info, PersonsExtraction pExtraction,
|
|
|
84 TextPart textPart) {
|
|
|
85 List<Paragraph> paragraphs = textPart.getParagraphs();
|
|
|
86 // find extraction object in list
|
|
|
87
|
|
|
88
|
|
|
89
|
|
|
90 List<ApplicantResult> results = new ArrayList<ApplicantResult>();
|
|
|
91 for (Paragraph para : paragraphs) {
|
|
|
92 Map<Integer, String> applicants = findApplicants(para.getParagraphText());
|
|
|
93 for (Integer pos : applicants.keySet()) {
|
|
|
94 // find out if person was already found
|
|
|
95 APerson person = null;
|
|
|
96 int posInText = textPart.getPositionInTextFromParagraph(para, pos);
|
|
|
97 for (APerson p : pExtraction.getPersons())
|
|
|
98 if (p.getStart() == posInText) {
|
|
|
99 person = p;
|
|
|
100 break;
|
|
|
101 }
|
|
|
102
|
|
|
103 if (person == null) {
|
|
|
104 person = new Person();
|
|
|
105 person.setStart(posInText);
|
|
|
106 pExtraction.getPersons().add(person);
|
|
|
107 }
|
|
|
108 person.setLength(applicants.get(pos).length());
|
|
|
109 person.setReferenceInText(applicants.get(pos));
|
|
|
110
|
|
|
111 ApplicantResult result = new ApplicantResult();
|
|
|
112 result.setPrediction(2.0);
|
|
|
113 result.setFinding(person);
|
|
|
114 result.setResult(pExtraction);
|
|
|
115 result.setTextInfo(info);
|
|
|
116 results.add(result);
|
|
|
117 }
|
|
|
118 }
|
|
|
119
|
|
|
120 return results;
|
|
|
121 }
|
|
|
122
|
|
|
123 public Map<Integer, String> findApplicants(String text) {
|
|
|
124 String paraText = text;
|
|
|
125 Pattern pattern = Pattern.compile(applicantRegex);
|
|
|
126 Matcher match = pattern.matcher(paraText);
|
|
|
127
|
|
|
128 Map<Integer, String> applicants = new HashMap<Integer, String>();
|
|
|
129 while (match.find()) {
|
|
|
130 String applicant = match.group(2);
|
|
|
131 int pos = match.start(2);
|
|
|
132 if (applicant.endsWith(","))
|
|
|
133 applicant = applicant.substring(0, applicant.length() - 1);
|
|
|
134 applicants.put(pos, applicant);
|
|
|
135 }
|
|
|
136
|
|
|
137 return applicants;
|
|
|
138 }
|
|
|
139 }
|
