装载人工干预词典
(self, filename, sep=None)
| 74 | self.ac.add_word(phrase) |
| 75 | |
| 76 | def load_customization(self, filename, sep=None): |
| 77 | """装载人工干预词典""" |
| 78 | self.ac = TriedTree() |
| 79 | with open(filename, 'r', encoding='utf8') as f: |
| 80 | for line in f: |
| 81 | if sep == None: |
| 82 | words = line.strip().split() |
| 83 | else: |
| 84 | sep = strdecode(sep) |
| 85 | words = line.strip().split(sep) |
| 86 | |
| 87 | if len(words) == 0: |
| 88 | continue |
| 89 | |
| 90 | phrase = "" |
| 91 | tags = [] |
| 92 | offset = [] |
| 93 | for word in words: |
| 94 | if word.rfind('/') < 1: |
| 95 | phrase += word |
| 96 | tags.append('') |
| 97 | else: |
| 98 | phrase += word[:word.rfind('/')] |
| 99 | tags.append(word[word.rfind('/') + 1:]) |
| 100 | offset.append(len(phrase)) |
| 101 | |
| 102 | if len(phrase) < 2 and tags[0] == '': |
| 103 | continue |
| 104 | |
| 105 | self.dictitem[phrase] = (tags, offset) |
| 106 | self.ac.add_word(phrase) |
| 107 | self.ac.make() |
| 108 | |
| 109 | def parse_customization(self, query, lac_tags): |
| 110 | """使用人工干预词典修正lac模型的输出""" |