1
0
mirror of https://github.com/fumiama/jieba.git synced 2026-06-27 07:30:32 +08:00

refactor analyse module

This commit is contained in:
Wang Bin
2015-05-04 16:39:37 +08:00
parent 500e6bd10e
commit 52fad00403
7 changed files with 187 additions and 236 deletions

View File

@@ -8,25 +8,26 @@ import (
var (
sentence = "此外公司拟对全资子公司吉林欧亚置业有限公司增资4.3亿元增资后吉林欧亚置业注册资本由7000万元增加到5亿元。吉林欧亚置业主要经营范围为房地产开发及百货零售等业务。目前在建吉林欧亚城市商业综合体项目。2013年实现营业收入0万元实现净利润-139.13万元。"
tagRanks = wordWeights{
wordWeight{Word: "吉林", Weight: 1.0},
wordWeight{Word: "欧亚", Weight: 0.87807810644},
wordWeight{Word: "置业", Weight: 0.562048250306},
wordWeight{Word: "实现", Weight: 0.520905743929},
wordWeight{Word: "收入", Weight: 0.384283870648},
wordWeight{Word: "增资", Weight: 0.360590945312},
wordWeight{Word: "子公司", Weight: 0.353131980904},
wordWeight{Word: "城市", Weight: 0.307509449283},
wordWeight{Word: "全资", Weight: 0.306324426665},
wordWeight{Word: "商业", Weight: 0.306138241063},
tagRanks = Segments{
Segment{text: "吉林", weight: 1.0},
Segment{text: "欧亚", weight: 0.87807810644},
Segment{text: "置业", weight: 0.562048250306},
Segment{text: "实现", weight: 0.520905743929},
Segment{text: "收入", weight: 0.384283870648},
Segment{text: "增资", weight: 0.360590945312},
Segment{text: "子公司", weight: 0.353131980904},
Segment{text: "城市", weight: 0.307509449283},
Segment{text: "全资", weight: 0.306324426665},
Segment{text: "商业", weight: 0.306138241063},
}
)
func TestTextRank(t *testing.T) {
tr, _ := NewTextRanker("../dict.txt")
var tr TextRanker
tr.LoadDictionary("../dict.txt")
results := tr.TextRank(sentence, 10)
for index, tw := range results {
if tw.Word != tagRanks[index].Word || math.Abs(tw.Weight-tagRanks[index].Weight) > 1e-6 {
if tw.text != tagRanks[index].text || math.Abs(tw.weight-tagRanks[index].weight) > 1e-6 {
t.Fatalf("%v != %v", tw, tagRanks[index])
}
}