fumiama/jieba
1
0
Fork 0
mirror of https://github.com/fumiama/jieba.git synced 2026-06-09 19:20:26 +08:00
Code Activity

small refactor the interface, use contructors instead of pointers for entry

Browse Source
This commit is contained in:
Wang Bin
2015-03-30 13:00:56 +08:00
parent 48a0bd390b
commit 51c63cb9ad
7 changed files with 274 additions and 283 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
analyse/idf.go
View File

@@ -11,7 +11,7 @@ type IDFLoader struct {
freqs []float64
}
func (l *IDFLoader) AddEntry(entry *jiebago.Entry) {
func (l *IDFLoader) AddEntry(entry jiebago.Entry) {
l.IDFFreq[entry.Word] = entry.Freq
l.freqs = append(l.freqs, entry.Freq)
}

2
analyse/stopwords.go
View File

@@ -40,7 +40,7 @@ type StopWordLoader struct {
stopWords map[string]int
}
func (s *StopWordLoader) AddEntry(entry *jiebago.Entry) {
func (s *StopWordLoader) AddEntry(entry jiebago.Entry) {
s.stopWords[entry.Word] = 1
}

12
dictionary.go
View File

@@ -1,21 +1,13 @@
package jiebago
type Pair struct {
type Entry struct {
Word string
Flag string
}
type Entry struct {
*Pair
Freq float64
}
func NewEntry() *Entry {
return &Entry{new(Pair), 0.0}
}
type DictLoader interface {
AddEntry(*Entry)
AddEntry(Entry)
}
type Cacher interface {

2
jieba.go
View File

@@ -55,7 +55,7 @@ type Jieba struct {
Freq map[string]float64
}
func (j *Jieba) AddEntry(entry *Entry) {
func (j *Jieba) AddEntry(entry Entry) {
j.Add(entry.Word, entry.Freq)
}

76
posseg/posseg.go
View File

@@ -16,8 +16,8 @@ var (
reSkipInternal = regexp.MustCompile(`(\r\n|\s)`)
)
type WordTag struct {
Word, Tag string
type Pair struct {
Word, Flag string
}
type Posseg struct {
@@ -25,7 +25,7 @@ type Posseg struct {
Flag map[string]string
}
func (p *Posseg) AddEntry(entry *jiebago.Entry) {
func (p *Posseg) AddEntry(entry jiebago.Entry) {
if len(entry.Flag) > 0 {
p.Flag[entry.Word] = strings.TrimSpace(entry.Flag)
}
@@ -49,8 +49,8 @@ func (p *Posseg) LoadUserDict(dictFilePath string) error {
return jiebago.LoadDict(p, dictFilePath, true)
}
func (p *Posseg) cutDetailInternal(sentence string) chan WordTag {
result := make(chan WordTag)
func (p *Posseg) cutDetailInternal(sentence string) chan Pair {
result := make(chan Pair)
go func() {
runes := []rune(sentence)
@@ -63,23 +63,23 @@ func (p *Posseg) cutDetailInternal(sentence string) chan WordTag {
case 'B':
begin = i
case 'E':
result <- WordTag{string(runes[begin : i+1]), string(pos[1:])}
result <- Pair{string(runes[begin : i+1]), string(pos[1:])}
next = i + 1
case 'S':
result <- WordTag{string(char), string(pos[1:])}
result <- Pair{string(char), string(pos[1:])}
next = i + 1
}
}
if next < len(runes) {
result <- WordTag{string(runes[next:]), string(posList[next][1:])}
result <- Pair{string(runes[next:]), string(posList[next][1:])}
}
close(result)
}()
return result
}
func (p *Posseg) cutDetail(sentence string) chan WordTag {
result := make(chan WordTag)
func (p *Posseg) cutDetail(sentence string) chan Pair {
result := make(chan Pair)
go func() {
for blk := range jiebago.RegexpSplit(reHanDetail, sentence) {
@@ -94,11 +94,11 @@ func (p *Posseg) cutDetail(sentence string) chan WordTag {
}
switch {
case reNum.MatchString(x):
result <- WordTag{x, "m"}
result <- Pair{x, "m"}
case reEng.MatchString(x):
result <- WordTag{x, "eng"}
result <- Pair{x, "eng"}
default:
result <- WordTag{x, "x"}
result <- Pair{x, "x"}
}
}
}
@@ -108,10 +108,10 @@ func (p *Posseg) cutDetail(sentence string) chan WordTag {
return result
}
type cutFunc func(sentence string) chan WordTag
type cutFunc func(sentence string) chan Pair
func (p *Posseg) cutDAG(sentence string) chan WordTag {
result := make(chan WordTag)
func (p *Posseg) cutDAG(sentence string) chan Pair {
result := make(chan Pair)
go func() {
dag := p.DAG(sentence)
@@ -130,9 +130,9 @@ func (p *Posseg) cutDAG(sentence string) chan WordTag {
if len(buf) == 1 {
sbuf := string(buf)
if tag, ok := p.Flag[sbuf]; ok {
result <- WordTag{sbuf, tag}
result <- Pair{sbuf, tag}
} else {
result <- WordTag{sbuf, "x"}
result <- Pair{sbuf, "x"}
}
buf = make([]rune, 0)
} else {
@@ -145,9 +145,9 @@ func (p *Posseg) cutDAG(sentence string) chan WordTag {
for _, elem := range buf {
selem := string(elem)
if tag, ok := p.Flag[selem]; ok {
result <- WordTag{string(elem), tag}
result <- Pair{string(elem), tag}
} else {
result <- WordTag{string(elem), "x"}
result <- Pair{string(elem), "x"}
}
}
@@ -157,9 +157,9 @@ func (p *Posseg) cutDAG(sentence string) chan WordTag {
}
sl_word := string(l_word)
if tag, ok := p.Flag[sl_word]; ok {
result <- WordTag{sl_word, tag}
result <- Pair{sl_word, tag}
} else {
result <- WordTag{sl_word, "x"}
result <- Pair{sl_word, "x"}
}
}
x = y
@@ -169,9 +169,9 @@ func (p *Posseg) cutDAG(sentence string) chan WordTag {
if len(buf) == 1 {
sbuf := string(buf)
if tag, ok := p.Flag[sbuf]; ok {
result <- WordTag{sbuf, tag}
result <- Pair{sbuf, tag}
} else {
result <- WordTag{sbuf, "x"}
result <- Pair{sbuf, "x"}
}
} else {
bufString := string(buf)
@@ -183,9 +183,9 @@ func (p *Posseg) cutDAG(sentence string) chan WordTag {
for _, elem := range buf {
selem := string(elem)
if tag, ok := p.Flag[selem]; ok {
result <- WordTag{selem, tag}
result <- Pair{selem, tag}
} else {
result <- WordTag{selem, "x"}
result <- Pair{selem, "x"}
}
}
}
@@ -196,8 +196,8 @@ func (p *Posseg) cutDAG(sentence string) chan WordTag {
return result
}
func (p *Posseg) cutDAGNoHMM(sentence string) chan WordTag {
result := make(chan WordTag)
func (p *Posseg) cutDAGNoHMM(sentence string) chan Pair {
result := make(chan Pair)
go func() {
dag := p.DAG(sentence)
@@ -218,20 +218,20 @@ func (p *Posseg) cutDAGNoHMM(sentence string) chan WordTag {
x = y
} else {
if len(buf) > 0 {
result <- WordTag{string(buf), "eng"}
result <- Pair{string(buf), "eng"}
buf = make([]rune, 0)
}
sl_word := string(l_word)
if tag, ok := p.Flag[sl_word]; ok {
result <- WordTag{sl_word, tag}
result <- Pair{sl_word, tag}
} else {
result <- WordTag{sl_word, "x"}
result <- Pair{sl_word, "x"}
}
x = y
}
}
if len(buf) > 0 {
result <- WordTag{string(buf), "eng"}
result <- Pair{string(buf), "eng"}
buf = make([]rune, 0)
}
close(result)
@@ -241,8 +241,8 @@ func (p *Posseg) cutDAGNoHMM(sentence string) chan WordTag {
// Tags the POS of each word after segmentation, using labels compatible with
// ictclas.
func (p *Posseg) Cut(sentence string, HMM bool) chan WordTag {
result := make(chan WordTag)
func (p *Posseg) Cut(sentence string, HMM bool) chan Pair {
result := make(chan Pair)
var cut cutFunc
if HMM {
cut = p.cutDAG
@@ -258,18 +258,18 @@ func (p *Posseg) Cut(sentence string, HMM bool) chan WordTag {
} else {
for x := range jiebago.RegexpSplit(reSkipInternal, blk) {
if reSkipInternal.MatchString(x) {
result <- WordTag{x, "x"}
result <- Pair{x, "x"}
} else {
for _, xx := range x {
s := string(xx)
switch {
case reNum.MatchString(s):
result <- WordTag{s, "m"}
result <- Pair{s, "m"}
case reEng.MatchString(x):
result <- WordTag{x, "eng"}
result <- Pair{x, "eng"}
break
default:
result <- WordTag{s, "x"}
result <- Pair{s, "x"}
}
}
}

460
posseg/posseg_test.go
View File

@@ -92,183 +92,183 @@ var (
"你认识那个和主席握手的的哥吗？他开一辆黑色的士。",
"枪杆子中出政权"}
defaultCutResult = [][]WordTag{[]WordTag{WordTag{"这", "r"}, WordTag{"是", "v"}, WordTag{"一个", "m"}, WordTag{"伸手不见五指", "i"}, WordTag{"的", "uj"}, WordTag{"黑夜", "n"}, WordTag{"。", "x"}, WordTag{"我", "r"}, WordTag{"叫", "v"}, WordTag{"孙悟空", "nr"}, WordTag{"，", "x"}, WordTag{"我", "r"}, WordTag{"爱", "v"}, WordTag{"北京", "ns"}, WordTag{"，", "x"}, WordTag{"我", "r"}, WordTag{"爱", "v"}, WordTag{"Python", "eng"}, WordTag{"和", "c"}, WordTag{"C++", "nz"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"不", "d"}, WordTag{"喜欢", "v"}, WordTag{"日本", "ns"}, WordTag{"和服", "nz"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"雷猴", "n"}, WordTag{"回归", "v"}, WordTag{"人间", "n"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"工信处", "n"}, WordTag{"女干事", "n"}, WordTag{"每月", "r"}, WordTag{"经过", "p"}, WordTag{"下属", "v"}, WordTag{"科室", "n"}, WordTag{"都", "d"}, WordTag{"要", "v"}, WordTag{"亲口", "n"}, WordTag{"交代", "n"}, WordTag{"24", "m"}, WordTag{"口", "n"}, WordTag{"交换机", "n"}, WordTag{"等", "u"}, WordTag{"技术性", "n"}, WordTag{"器件", "n"}, WordTag{"的", "uj"}, WordTag{"安装", "v"}, WordTag{"工作", "vn"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"需要", "v"}, WordTag{"廉租房", "n"}},
[]WordTag{WordTag{"永和", "nz"}, WordTag{"服装", "vn"}, WordTag{"饰品", "n"}, WordTag{"有限公司", "n"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"爱", "v"}, WordTag{"北京", "ns"}, WordTag{"天安门", "ns"}},
[]WordTag{WordTag{"abc", "eng"}},
[]WordTag{WordTag{"隐", "n"}, WordTag{"马尔可夫", "nr"}},
[]WordTag{WordTag{"雷猴", "n"}, WordTag{"是", "v"}, WordTag{"个", "q"}, WordTag{"好", "a"}, WordTag{"网站", "n"}},
[]WordTag{WordTag{"“", "x"}, WordTag{"Microsoft", "eng"}, WordTag{"”", "x"}, WordTag{"一", "m"}, WordTag{"词", "n"}, WordTag{"由", "p"}, WordTag{"“", "x"}, WordTag{"MICROcomputer", "eng"}, WordTag{"（", "x"}, WordTag{"微型", "b"}, WordTag{"计算机", "n"}, WordTag{"）", "x"}, WordTag{"”", "x"}, WordTag{"和", "c"}, WordTag{"“", "x"}, WordTag{"SOFTware", "eng"}, WordTag{"（", "x"}, WordTag{"软件", "n"}, WordTag{"）", "x"}, WordTag{"”", "x"}, WordTag{"两", "m"}, WordTag{"部分", "n"}, WordTag{"组成", "v"}},
[]WordTag{WordTag{"草泥马", "n"}, WordTag{"和", "c"}, WordTag{"欺实", "v"}, WordTag{"马", "n"}, WordTag{"是", "v"}, WordTag{"今年", "t"}, WordTag{"的", "uj"}, WordTag{"流行", "v"}, WordTag{"词汇", "n"}},
[]WordTag{WordTag{"伊藤", "nr"}, WordTag{"洋华堂", "n"}, WordTag{"总府", "n"}, WordTag{"店", "n"}},
[]WordTag{WordTag{"中国科学院计算技术研究所", "nt"}},
[]WordTag{WordTag{"罗密欧", "nr"}, WordTag{"与", "p"}, WordTag{"朱丽叶", "nr"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"购买", "v"}, WordTag{"了", "ul"}, WordTag{"道具", "n"}, WordTag{"和", "c"}, WordTag{"服装", "vn"}},
[]WordTag{WordTag{"PS", "eng"}, WordTag{":", "x"}, WordTag{" ", "x"}, WordTag{"我", "r"}, WordTag{"觉得", "v"}, WordTag{"开源", "n"}, WordTag{"有", "v"}, WordTag{"一个", "m"}, WordTag{"好处", "d"}, WordTag{"，", "x"}, WordTag{"就是", "d"}, WordTag{"能够", "v"}, WordTag{"敦促", "v"}, WordTag{"自己", "r"}, WordTag{"不断改进", "l"}, WordTag{"，", "x"}, WordTag{"避免", "v"}, WordTag{"敞", "v"}, WordTag{"帚", "ng"}, WordTag{"自珍", "b"}},
[]WordTag{WordTag{"湖北省", "ns"}, WordTag{"石首市", "ns"}},
[]WordTag{WordTag{"湖北省", "ns"}, WordTag{"十堰市", "ns"}},
[]WordTag{WordTag{"总经理", "n"}, WordTag{"完成", "v"}, WordTag{"了", "ul"}, WordTag{"这件", "mq"}, WordTag{"事情", "n"}},
[]WordTag{WordTag{"电脑", "n"}, WordTag{"修好", "v"}, WordTag{"了", "ul"}},
[]WordTag{WordTag{"做好", "v"}, WordTag{"了", "ul"}, WordTag{"这件", "mq"}, WordTag{"事情", "n"}, WordTag{"就", "d"}, WordTag{"一了百了", "l"}, WordTag{"了", "ul"}},
[]WordTag{WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"我们", "r"}, WordTag{"买", "v"}, WordTag{"了", "ul"}, WordTag{"一个", "m"}, WordTag{"美的", "nr"}, WordTag{"空调", "n"}},
[]WordTag{WordTag{"线程", "n"}, WordTag{"初始化", "l"}, WordTag{"时", "n"}, WordTag{"我们", "r"}, WordTag{"要", "v"}, WordTag{"注意", "v"}},
[]WordTag{WordTag{"一个", "m"}, WordTag{"分子", "n"}, WordTag{"是", "v"}, WordTag{"由", "p"}, WordTag{"好多", "m"}, WordTag{"原子", "n"}, WordTag{"组织", "v"}, WordTag{"成", "v"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"祝", "v"}, WordTag{"你", "r"}, WordTag{"马到功成", "i"}},
[]WordTag{WordTag{"他", "r"}, WordTag{"掉", "v"}, WordTag{"进", "v"}, WordTag{"了", "ul"}, WordTag{"无底洞", "ns"}, WordTag{"里", "f"}},
[]WordTag{WordTag{"中国", "ns"}, WordTag{"的", "uj"}, WordTag{"首都", "d"}, WordTag{"是", "v"}, WordTag{"北京", "ns"}},
[]WordTag{WordTag{"孙君意", "nr"}},
[]WordTag{WordTag{"外交部", "nt"}, WordTag{"发言人", "l"}, WordTag{"马朝旭", "nr"}},
[]WordTag{WordTag{"领导人", "n"}, WordTag{"会议", "n"}, WordTag{"和", "c"}, WordTag{"第四届", "m"}, WordTag{"东亚", "ns"}, WordTag{"峰会", "n"}},
[]WordTag{WordTag{"在", "p"}, WordTag{"过去", "t"}, WordTag{"的", "uj"}, WordTag{"这", "r"}, WordTag{"五年", "t"}},
[]WordTag{WordTag{"还", "d"}, WordTag{"需要", "v"}, WordTag{"很", "d"}, WordTag{"长", "a"}, WordTag{"的", "uj"}, WordTag{"路", "n"}, WordTag{"要", "v"}, WordTag{"走", "v"}},
[]WordTag{WordTag{"60", "m"}, WordTag{"周年", "t"}, WordTag{"首都", "d"}, WordTag{"阅兵", "v"}},
[]WordTag{WordTag{"你好", "l"}, WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"买", "v"}, WordTag{"水果", "n"}, WordTag{"然后", "c"}, WordTag{"来", "v"}, WordTag{"世博园", "nr"}},
[]WordTag{WordTag{"买", "v"}, WordTag{"水果", "n"}, WordTag{"然后", "c"}, WordTag{"去", "v"}, WordTag{"世博园", "nr"}},
[]WordTag{WordTag{"但是", "c"}, WordTag{"后来", "t"}, WordTag{"我", "r"}, WordTag{"才", "d"}, WordTag{"知道", "v"}, WordTag{"你", "r"}, WordTag{"是", "v"}, WordTag{"对", "p"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"存在", "v"}, WordTag{"即", "v"}, WordTag{"合理", "vn"}},
[]WordTag{WordTag{"的的", "u"}, WordTag{"的的", "u"}, WordTag{"的", "uj"}, WordTag{"在的", "u"}, WordTag{"的的", "u"}, WordTag{"的", "uj"}, WordTag{"就", "d"}, WordTag{"以", "p"}, WordTag{"和和", "nz"}, WordTag{"和", "c"}},
[]WordTag{WordTag{"I", "x"}, WordTag{" ", "x"}, WordTag{"love", "eng"}, WordTag{"你", "r"}, WordTag{"，", "x"}, WordTag{"不以为耻", "i"}, WordTag{"，", "x"}, WordTag{"反", "zg"}, WordTag{"以为", "c"}, WordTag{"rong", "eng"}},
[]WordTag{WordTag{"因", "p"}},
[]WordTag{},
[]WordTag{WordTag{"hello", "eng"}, WordTag{"你好", "l"}, WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"很好", "a"}, WordTag{"但", "c"}, WordTag{"主要", "b"}, WordTag{"是", "v"}, WordTag{"基于", "p"}, WordTag{"网页", "n"}, WordTag{"形式", "n"}},
[]WordTag{WordTag{"hello", "eng"}, WordTag{"你好", "l"}, WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"为什么", "r"}, WordTag{"我", "r"}, WordTag{"不能", "v"}, WordTag{"拥有", "v"}, WordTag{"想要", "v"}, WordTag{"的", "uj"}, WordTag{"生活", "vn"}},
[]WordTag{WordTag{"后来", "t"}, WordTag{"我", "r"}, WordTag{"才", "d"}},
[]WordTag{WordTag{"此次", "r"}, WordTag{"来", "v"}, WordTag{"中国", "ns"}, WordTag{"是", "v"}, WordTag{"为了", "p"}},
[]WordTag{WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{",", "x"}, WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{"其实", "d"}, WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{"好人", "n"}, WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{"是因为", "c"}, WordTag{"和", "c"}, WordTag{"国家", "n"}},
[]WordTag{WordTag{"老年", "t"}, WordTag{"搜索", "v"}, WordTag{"还", "d"}, WordTag{"支持", "v"}},
[]WordTag{WordTag{"干脆", "d"}, WordTag{"就", "d"}, WordTag{"把", "p"}, WordTag{"那部", "r"}, WordTag{"蒙人", "n"}, WordTag{"的", "uj"}, WordTag{"闲法", "n"}, WordTag{"给", "p"}, WordTag{"废", "v"}, WordTag{"了", "ul"}, WordTag{"拉倒", "v"}, WordTag{"！", "x"}, WordTag{"RT", "eng"}, WordTag{" ", "x"}, WordTag{"@", "x"}, WordTag{"laoshipukong", "eng"}, WordTag{" ", "x"}, WordTag{":", "x"}, WordTag{" ", "x"}, WordTag{"27", "m"}, WordTag{"日", "m"}, WordTag{"，", "x"}, WordTag{"全国人大常委会", "nt"}, WordTag{"第三次", "m"}, WordTag{"审议", "v"}, WordTag{"侵权", "v"}, WordTag{"责任法", "n"}, WordTag{"草案", "n"}, WordTag{"，", "x"}, WordTag{"删除", "v"}, WordTag{"了", "ul"}, WordTag{"有关", "vn"}, WordTag{"医疗", "n"}, WordTag{"损害", "v"}, WordTag{"责任", "n"}, WordTag{"“", "x"}, WordTag{"举证", "v"}, WordTag{"倒置", "v"}, WordTag{"”", "x"}, WordTag{"的", "uj"}, WordTag{"规定", "n"}, WordTag{"。", "x"}, WordTag{"在", "p"}, WordTag{"医患", "n"}, WordTag{"纠纷", "n"}, WordTag{"中本", "ns"}, WordTag{"已", "d"}, WordTag{"处于", "v"}, WordTag{"弱势", "n"}, WordTag{"地位", "n"}, WordTag{"的", "uj"}, WordTag{"消费者", "n"}, WordTag{"由此", "c"}, WordTag{"将", "d"}, WordTag{"陷入", "v"}, WordTag{"万劫不复", "i"}, WordTag{"的", "uj"}, WordTag{"境地", "s"}, WordTag{"。", "x"}, WordTag{" ", "x"}},
[]WordTag{WordTag{"大", "a"}},
[]WordTag{},
[]WordTag{WordTag{"他", "r"}, WordTag{"说", "v"}, WordTag{"的", "uj"}, WordTag{"确实", "ad"}, WordTag{"在", "p"}, WordTag{"理", "n"}},
[]WordTag{WordTag{"长春", "ns"}, WordTag{"市长", "n"}, WordTag{"春节", "t"}, WordTag{"讲话", "n"}},
[]WordTag{WordTag{"结婚", "v"}, WordTag{"的", "uj"}, WordTag{"和", "c"}, WordTag{"尚未", "d"}, WordTag{"结婚", "v"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"结合", "v"}, WordTag{"成", "n"}, WordTag{"分子", "n"}, WordTag{"时", "n"}},
[]WordTag{WordTag{"旅游", "vn"}, WordTag{"和", "c"}, WordTag{"服务", "vn"}, WordTag{"是", "v"}, WordTag{"最好", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"这件", "mq"}, WordTag{"事情", "n"}, WordTag{"的确", "d"}, WordTag{"是", "v"}, WordTag{"我", "r"}, WordTag{"的", "uj"}, WordTag{"错", "n"}},
[]WordTag{WordTag{"供", "v"}, WordTag{"大家", "n"}, WordTag{"参考", "v"}, WordTag{"指正", "v"}},
[]WordTag{WordTag{"哈尔滨", "ns"}, WordTag{"政府", "n"}, WordTag{"公布", "v"}, WordTag{"塌", "v"}, WordTag{"桥", "n"}, WordTag{"原因", "n"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"在", "p"}, WordTag{"机场", "n"}, WordTag{"入口处", "i"}},
[]WordTag{WordTag{"邢永臣", "nr"}, WordTag{"摄影", "n"}, WordTag{"报道", "v"}},
[]WordTag{WordTag{"BP", "eng"}, WordTag{"神经网络", "n"}, WordTag{"如何", "r"}, WordTag{"训练", "vn"}, WordTag{"才能", "v"}, WordTag{"在", "p"}, WordTag{"分类", "n"}, WordTag{"时", "n"}, WordTag{"增加", "v"}, WordTag{"区分度", "n"}, WordTag{"？", "x"}},
[]WordTag{WordTag{"南京市", "ns"}, WordTag{"长江大桥", "ns"}},
[]WordTag{WordTag{"应", "v"}, WordTag{"一些", "m"}, WordTag{"使用者", "n"}, WordTag{"的", "uj"}, WordTag{"建议", "n"}, WordTag{"，", "x"}, WordTag{"也", "d"}, WordTag{"为了", "p"}, WordTag{"便于", "v"}, WordTag{"利用", "n"}, WordTag{"NiuTrans", "eng"}, WordTag{"用于", "v"}, WordTag{"SMT", "eng"}, WordTag{"研究", "vn"}},
[]WordTag{WordTag{"长春市", "ns"}, WordTag{"长春", "ns"}, WordTag{"药店", "n"}},
[]WordTag{WordTag{"邓颖超", "nr"}, WordTag{"生前", "t"}, WordTag{"最", "d"}, WordTag{"喜欢", "v"}, WordTag{"的", "uj"}, WordTag{"衣服", "n"}},
[]WordTag{WordTag{"胡锦涛", "nr"}, WordTag{"是", "v"}, WordTag{"热爱", "a"}, WordTag{"世界", "n"}, WordTag{"和平", "nz"}, WordTag{"的", "uj"}, WordTag{"政治局", "n"}, WordTag{"常委", "j"}},
[]WordTag{WordTag{"程序员", "n"}, WordTag{"祝", "v"}, WordTag{"海林", "nz"}, WordTag{"和", "c"}, WordTag{"朱会震", "nr"}, WordTag{"是", "v"}, WordTag{"在", "p"}, WordTag{"孙健", "nr"}, WordTag{"的", "uj"}, WordTag{"左面", "f"}, WordTag{"和", "c"}, WordTag{"右面", "f"}, WordTag{",", "x"}, WordTag{" ", "x"}, WordTag{"范凯", "nr"}, WordTag{"在", "p"}, WordTag{"最", "a"}, WordTag{"右面", "f"}, WordTag{".", "m"}, WordTag{"再往", "d"}, WordTag{"左", "f"}, WordTag{"是", "v"}, WordTag{"李松洪", "nr"}},
[]WordTag{WordTag{"一次性", "d"}, WordTag{"交", "v"}, WordTag{"多少", "m"}, WordTag{"钱", "n"}},
[]WordTag{WordTag{"两块", "m"}, WordTag{"五", "m"}, WordTag{"一套", "m"}, WordTag{"，", "x"}, WordTag{"三块", "m"}, WordTag{"八", "m"}, WordTag{"一斤", "m"}, WordTag{"，", "x"}, WordTag{"四块", "m"}, WordTag{"七", "m"}, WordTag{"一本", "m"}, WordTag{"，", "x"}, WordTag{"五块", "m"}, WordTag{"六", "m"}, WordTag{"一条", "m"}},
[]WordTag{WordTag{"小", "a"}, WordTag{"和尚", "nr"}, WordTag{"留", "v"}, WordTag{"了", "ul"}, WordTag{"一个", "m"}, WordTag{"像", "v"}, WordTag{"大", "a"}, WordTag{"和尚", "nr"}, WordTag{"一样", "r"}, WordTag{"的", "uj"}, WordTag{"和尚头", "nr"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"是", "v"}, WordTag{"中华人民共和国", "ns"}, WordTag{"公民", "n"}, WordTag{";", "x"}, WordTag{"我", "r"}, WordTag{"爸爸", "n"}, WordTag{"是", "v"}, WordTag{"共和党", "nt"}, WordTag{"党员", "n"}, WordTag{";", "x"}, WordTag{" ", "x"}, WordTag{"地铁", "n"}, WordTag{"和平门", "ns"}, WordTag{"站", "v"}},
[]WordTag{WordTag{"张晓梅", "nr"}, WordTag{"去", "v"}, WordTag{"人民", "n"}, WordTag{"医院", "n"}, WordTag{"做", "v"}, WordTag{"了", "ul"}, WordTag{"个", "q"}, WordTag{"B超", "n"}, WordTag{"然后", "c"}, WordTag{"去", "v"}, WordTag{"买", "v"}, WordTag{"了", "ul"}, WordTag{"件", "q"}, WordTag{"T恤", "n"}},
[]WordTag{WordTag{"AT&T", "nz"}, WordTag{"是", "v"}, WordTag{"一件", "m"}, WordTag{"不错", "a"}, WordTag{"的", "uj"}, WordTag{"公司", "n"}, WordTag{"，", "x"}, WordTag{"给", "p"}, WordTag{"你", "r"}, WordTag{"发", "v"}, WordTag{"offer", "eng"}, WordTag{"了", "ul"}, WordTag{"吗", "y"}, WordTag{"？", "x"}},
[]WordTag{WordTag{"C++", "nz"}, WordTag{"和", "c"}, WordTag{"c#", "nz"}, WordTag{"是", "v"}, WordTag{"什么", "r"}, WordTag{"关系", "n"}, WordTag{"？", "x"}, WordTag{"11", "m"}, WordTag{"+", "x"}, WordTag{"122", "m"}, WordTag{"=", "x"}, WordTag{"133", "m"}, WordTag{"，", "x"}, WordTag{"是", "v"}, WordTag{"吗", "y"}, WordTag{"？", "x"}, WordTag{"PI", "eng"}, WordTag{"=", "x"}, WordTag{"3.14159", "m"}},
[]WordTag{WordTag{"你", "r"}, WordTag{"认识", "v"}, WordTag{"那个", "r"}, WordTag{"和", "c"}, WordTag{"主席", "n"}, WordTag{"握手", "v"}, WordTag{"的", "uj"}, WordTag{"的哥", "n"}, WordTag{"吗", "y"}, WordTag{"？", "x"}, WordTag{"他", "r"}, WordTag{"开", "v"}, WordTag{"一辆", "m"}, WordTag{"黑色", "n"}, WordTag{"的士", "n"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"枪杆子", "n"}, WordTag{"中", "f"}, WordTag{"出", "v"}, WordTag{"政权", "n"}},
defaultCutResult = [][]Pair{[]Pair{Pair{"这", "r"}, Pair{"是", "v"}, Pair{"一个", "m"}, Pair{"伸手不见五指", "i"}, Pair{"的", "uj"}, Pair{"黑夜", "n"}, Pair{"。", "x"}, Pair{"我", "r"}, Pair{"叫", "v"}, Pair{"孙悟空", "nr"}, Pair{"，", "x"}, Pair{"我", "r"}, Pair{"爱", "v"}, Pair{"北京", "ns"}, Pair{"，", "x"}, Pair{"我", "r"}, Pair{"爱", "v"}, Pair{"Python", "eng"}, Pair{"和", "c"}, Pair{"C++", "nz"}, Pair{"。", "x"}},
[]Pair{Pair{"我", "r"}, Pair{"不", "d"}, Pair{"喜欢", "v"}, Pair{"日本", "ns"}, Pair{"和服", "nz"}, Pair{"。", "x"}},
[]Pair{Pair{"雷猴", "n"}, Pair{"回归", "v"}, Pair{"人间", "n"}, Pair{"。", "x"}},
[]Pair{Pair{"工信处", "n"}, Pair{"女干事", "n"}, Pair{"每月", "r"}, Pair{"经过", "p"}, Pair{"下属", "v"}, Pair{"科室", "n"}, Pair{"都", "d"}, Pair{"要", "v"}, Pair{"亲口", "n"}, Pair{"交代", "n"}, Pair{"24", "m"}, Pair{"口", "n"}, Pair{"交换机", "n"}, Pair{"等", "u"}, Pair{"技术性", "n"}, Pair{"器件", "n"}, Pair{"的", "uj"}, Pair{"安装", "v"}, Pair{"工作", "vn"}},
[]Pair{Pair{"我", "r"}, Pair{"需要", "v"}, Pair{"廉租房", "n"}},
[]Pair{Pair{"永和", "nz"}, Pair{"服装", "vn"}, Pair{"饰品", "n"}, Pair{"有限公司", "n"}},
[]Pair{Pair{"我", "r"}, Pair{"爱", "v"}, Pair{"北京", "ns"}, Pair{"天安门", "ns"}},
[]Pair{Pair{"abc", "eng"}},
[]Pair{Pair{"隐", "n"}, Pair{"马尔可夫", "nr"}},
[]Pair{Pair{"雷猴", "n"}, Pair{"是", "v"}, Pair{"个", "q"}, Pair{"好", "a"}, Pair{"网站", "n"}},
[]Pair{Pair{"“", "x"}, Pair{"Microsoft", "eng"}, Pair{"”", "x"}, Pair{"一", "m"}, Pair{"词", "n"}, Pair{"由", "p"}, Pair{"“", "x"}, Pair{"MICROcomputer", "eng"}, Pair{"（", "x"}, Pair{"微型", "b"}, Pair{"计算机", "n"}, Pair{"）", "x"}, Pair{"”", "x"}, Pair{"和", "c"}, Pair{"“", "x"}, Pair{"SOFTware", "eng"}, Pair{"（", "x"}, Pair{"软件", "n"}, Pair{"）", "x"}, Pair{"”", "x"}, Pair{"两", "m"}, Pair{"部分", "n"}, Pair{"组成", "v"}},
[]Pair{Pair{"草泥马", "n"}, Pair{"和", "c"}, Pair{"欺实", "v"}, Pair{"马", "n"}, Pair{"是", "v"}, Pair{"今年", "t"}, Pair{"的", "uj"}, Pair{"流行", "v"}, Pair{"词汇", "n"}},
[]Pair{Pair{"伊藤", "nr"}, Pair{"洋华堂", "n"}, Pair{"总府", "n"}, Pair{"店", "n"}},
[]Pair{Pair{"中国科学院计算技术研究所", "nt"}},
[]Pair{Pair{"罗密欧", "nr"}, Pair{"与", "p"}, Pair{"朱丽叶", "nr"}},
[]Pair{Pair{"我", "r"}, Pair{"购买", "v"}, Pair{"了", "ul"}, Pair{"道具", "n"}, Pair{"和", "c"}, Pair{"服装", "vn"}},
[]Pair{Pair{"PS", "eng"}, Pair{":", "x"}, Pair{" ", "x"}, Pair{"我", "r"}, Pair{"觉得", "v"}, Pair{"开源", "n"}, Pair{"有", "v"}, Pair{"一个", "m"}, Pair{"好处", "d"}, Pair{"，", "x"}, Pair{"就是", "d"}, Pair{"能够", "v"}, Pair{"敦促", "v"}, Pair{"自己", "r"}, Pair{"不断改进", "l"}, Pair{"，", "x"}, Pair{"避免", "v"}, Pair{"敞", "v"}, Pair{"帚", "ng"}, Pair{"自珍", "b"}},
[]Pair{Pair{"湖北省", "ns"}, Pair{"石首市", "ns"}},
[]Pair{Pair{"湖北省", "ns"}, Pair{"十堰市", "ns"}},
[]Pair{Pair{"总经理", "n"}, Pair{"完成", "v"}, Pair{"了", "ul"}, Pair{"这件", "mq"}, Pair{"事情", "n"}},
[]Pair{Pair{"电脑", "n"}, Pair{"修好", "v"}, Pair{"了", "ul"}},
[]Pair{Pair{"做好", "v"}, Pair{"了", "ul"}, Pair{"这件", "mq"}, Pair{"事情", "n"}, Pair{"就", "d"}, Pair{"一了百了", "l"}, Pair{"了", "ul"}},
[]Pair{Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"我们", "r"}, Pair{"买", "v"}, Pair{"了", "ul"}, Pair{"一个", "m"}, Pair{"美的", "nr"}, Pair{"空调", "n"}},
[]Pair{Pair{"线程", "n"}, Pair{"初始化", "l"}, Pair{"时", "n"}, Pair{"我们", "r"}, Pair{"要", "v"}, Pair{"注意", "v"}},
[]Pair{Pair{"一个", "m"}, Pair{"分子", "n"}, Pair{"是", "v"}, Pair{"由", "p"}, Pair{"好多", "m"}, Pair{"原子", "n"}, Pair{"组织", "v"}, Pair{"成", "v"}, Pair{"的", "uj"}},
[]Pair{Pair{"祝", "v"}, Pair{"你", "r"}, Pair{"马到功成", "i"}},
[]Pair{Pair{"他", "r"}, Pair{"掉", "v"}, Pair{"进", "v"}, Pair{"了", "ul"}, Pair{"无底洞", "ns"}, Pair{"里", "f"}},
[]Pair{Pair{"中国", "ns"}, Pair{"的", "uj"}, Pair{"首都", "d"}, Pair{"是", "v"}, Pair{"北京", "ns"}},
[]Pair{Pair{"孙君意", "nr"}},
[]Pair{Pair{"外交部", "nt"}, Pair{"发言人", "l"}, Pair{"马朝旭", "nr"}},
[]Pair{Pair{"领导人", "n"}, Pair{"会议", "n"}, Pair{"和", "c"}, Pair{"第四届", "m"}, Pair{"东亚", "ns"}, Pair{"峰会", "n"}},
[]Pair{Pair{"在", "p"}, Pair{"过去", "t"}, Pair{"的", "uj"}, Pair{"这", "r"}, Pair{"五年", "t"}},
[]Pair{Pair{"还", "d"}, Pair{"需要", "v"}, Pair{"很", "d"}, Pair{"长", "a"}, Pair{"的", "uj"}, Pair{"路", "n"}, Pair{"要", "v"}, Pair{"走", "v"}},
[]Pair{Pair{"60", "m"}, Pair{"周年", "t"}, Pair{"首都", "d"}, Pair{"阅兵", "v"}},
[]Pair{Pair{"你好", "l"}, Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"买", "v"}, Pair{"水果", "n"}, Pair{"然后", "c"}, Pair{"来", "v"}, Pair{"世博园", "nr"}},
[]Pair{Pair{"买", "v"}, Pair{"水果", "n"}, Pair{"然后", "c"}, Pair{"去", "v"}, Pair{"世博园", "nr"}},
[]Pair{Pair{"但是", "c"}, Pair{"后来", "t"}, Pair{"我", "r"}, Pair{"才", "d"}, Pair{"知道", "v"}, Pair{"你", "r"}, Pair{"是", "v"}, Pair{"对", "p"}, Pair{"的", "uj"}},
[]Pair{Pair{"存在", "v"}, Pair{"即", "v"}, Pair{"合理", "vn"}},
[]Pair{Pair{"的的", "u"}, Pair{"的的", "u"}, Pair{"的", "uj"}, Pair{"在的", "u"}, Pair{"的的", "u"}, Pair{"的", "uj"}, Pair{"就", "d"}, Pair{"以", "p"}, Pair{"和和", "nz"}, Pair{"和", "c"}},
[]Pair{Pair{"I", "x"}, Pair{" ", "x"}, Pair{"love", "eng"}, Pair{"你", "r"}, Pair{"，", "x"}, Pair{"不以为耻", "i"}, Pair{"，", "x"}, Pair{"反", "zg"}, Pair{"以为", "c"}, Pair{"rong", "eng"}},
[]Pair{Pair{"因", "p"}},
[]Pair{},
[]Pair{Pair{"hello", "eng"}, Pair{"你好", "l"}, Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"很好", "a"}, Pair{"但", "c"}, Pair{"主要", "b"}, Pair{"是", "v"}, Pair{"基于", "p"}, Pair{"网页", "n"}, Pair{"形式", "n"}},
[]Pair{Pair{"hello", "eng"}, Pair{"你好", "l"}, Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"为什么", "r"}, Pair{"我", "r"}, Pair{"不能", "v"}, Pair{"拥有", "v"}, Pair{"想要", "v"}, Pair{"的", "uj"}, Pair{"生活", "vn"}},
[]Pair{Pair{"后来", "t"}, Pair{"我", "r"}, Pair{"才", "d"}},
[]Pair{Pair{"此次", "r"}, Pair{"来", "v"}, Pair{"中国", "ns"}, Pair{"是", "v"}, Pair{"为了", "p"}},
[]Pair{Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{",", "x"}, Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{"其实", "d"}, Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{"好人", "n"}, Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{"是因为", "c"}, Pair{"和", "c"}, Pair{"国家", "n"}},
[]Pair{Pair{"老年", "t"}, Pair{"搜索", "v"}, Pair{"还", "d"}, Pair{"支持", "v"}},
[]Pair{Pair{"干脆", "d"}, Pair{"就", "d"}, Pair{"把", "p"}, Pair{"那部", "r"}, Pair{"蒙人", "n"}, Pair{"的", "uj"}, Pair{"闲法", "n"}, Pair{"给", "p"}, Pair{"废", "v"}, Pair{"了", "ul"}, Pair{"拉倒", "v"}, Pair{"！", "x"}, Pair{"RT", "eng"}, Pair{" ", "x"}, Pair{"@", "x"}, Pair{"laoshipukong", "eng"}, Pair{" ", "x"}, Pair{":", "x"}, Pair{" ", "x"}, Pair{"27", "m"}, Pair{"日", "m"}, Pair{"，", "x"}, Pair{"全国人大常委会", "nt"}, Pair{"第三次", "m"}, Pair{"审议", "v"}, Pair{"侵权", "v"}, Pair{"责任法", "n"}, Pair{"草案", "n"}, Pair{"，", "x"}, Pair{"删除", "v"}, Pair{"了", "ul"}, Pair{"有关", "vn"}, Pair{"医疗", "n"}, Pair{"损害", "v"}, Pair{"责任", "n"}, Pair{"“", "x"}, Pair{"举证", "v"}, Pair{"倒置", "v"}, Pair{"”", "x"}, Pair{"的", "uj"}, Pair{"规定", "n"}, Pair{"。", "x"}, Pair{"在", "p"}, Pair{"医患", "n"}, Pair{"纠纷", "n"}, Pair{"中本", "ns"}, Pair{"已", "d"}, Pair{"处于", "v"}, Pair{"弱势", "n"}, Pair{"地位", "n"}, Pair{"的", "uj"}, Pair{"消费者", "n"}, Pair{"由此", "c"}, Pair{"将", "d"}, Pair{"陷入", "v"}, Pair{"万劫不复", "i"}, Pair{"的", "uj"}, Pair{"境地", "s"}, Pair{"。", "x"}, Pair{" ", "x"}},
[]Pair{Pair{"大", "a"}},
[]Pair{},
[]Pair{Pair{"他", "r"}, Pair{"说", "v"}, Pair{"的", "uj"}, Pair{"确实", "ad"}, Pair{"在", "p"}, Pair{"理", "n"}},
[]Pair{Pair{"长春", "ns"}, Pair{"市长", "n"}, Pair{"春节", "t"}, Pair{"讲话", "n"}},
[]Pair{Pair{"结婚", "v"}, Pair{"的", "uj"}, Pair{"和", "c"}, Pair{"尚未", "d"}, Pair{"结婚", "v"}, Pair{"的", "uj"}},
[]Pair{Pair{"结合", "v"}, Pair{"成", "n"}, Pair{"分子", "n"}, Pair{"时", "n"}},
[]Pair{Pair{"旅游", "vn"}, Pair{"和", "c"}, Pair{"服务", "vn"}, Pair{"是", "v"}, Pair{"最好", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"这件", "mq"}, Pair{"事情", "n"}, Pair{"的确", "d"}, Pair{"是", "v"}, Pair{"我", "r"}, Pair{"的", "uj"}, Pair{"错", "n"}},
[]Pair{Pair{"供", "v"}, Pair{"大家", "n"}, Pair{"参考", "v"}, Pair{"指正", "v"}},
[]Pair{Pair{"哈尔滨", "ns"}, Pair{"政府", "n"}, Pair{"公布", "v"}, Pair{"塌", "v"}, Pair{"桥", "n"}, Pair{"原因", "n"}},
[]Pair{Pair{"我", "r"}, Pair{"在", "p"}, Pair{"机场", "n"}, Pair{"入口处", "i"}},
[]Pair{Pair{"邢永臣", "nr"}, Pair{"摄影", "n"}, Pair{"报道", "v"}},
[]Pair{Pair{"BP", "eng"}, Pair{"神经网络", "n"}, Pair{"如何", "r"}, Pair{"训练", "vn"}, Pair{"才能", "v"}, Pair{"在", "p"}, Pair{"分类", "n"}, Pair{"时", "n"}, Pair{"增加", "v"}, Pair{"区分度", "n"}, Pair{"？", "x"}},
[]Pair{Pair{"南京市", "ns"}, Pair{"长江大桥", "ns"}},
[]Pair{Pair{"应", "v"}, Pair{"一些", "m"}, Pair{"使用者", "n"}, Pair{"的", "uj"}, Pair{"建议", "n"}, Pair{"，", "x"}, Pair{"也", "d"}, Pair{"为了", "p"}, Pair{"便于", "v"}, Pair{"利用", "n"}, Pair{"NiuTrans", "eng"}, Pair{"用于", "v"}, Pair{"SMT", "eng"}, Pair{"研究", "vn"}},
[]Pair{Pair{"长春市", "ns"}, Pair{"长春", "ns"}, Pair{"药店", "n"}},
[]Pair{Pair{"邓颖超", "nr"}, Pair{"生前", "t"}, Pair{"最", "d"}, Pair{"喜欢", "v"}, Pair{"的", "uj"}, Pair{"衣服", "n"}},
[]Pair{Pair{"胡锦涛", "nr"}, Pair{"是", "v"}, Pair{"热爱", "a"}, Pair{"世界", "n"}, Pair{"和平", "nz"}, Pair{"的", "uj"}, Pair{"政治局", "n"}, Pair{"常委", "j"}},
[]Pair{Pair{"程序员", "n"}, Pair{"祝", "v"}, Pair{"海林", "nz"}, Pair{"和", "c"}, Pair{"朱会震", "nr"}, Pair{"是", "v"}, Pair{"在", "p"}, Pair{"孙健", "nr"}, Pair{"的", "uj"}, Pair{"左面", "f"}, Pair{"和", "c"}, Pair{"右面", "f"}, Pair{",", "x"}, Pair{" ", "x"}, Pair{"范凯", "nr"}, Pair{"在", "p"}, Pair{"最", "a"}, Pair{"右面", "f"}, Pair{".", "m"}, Pair{"再往", "d"}, Pair{"左", "f"}, Pair{"是", "v"}, Pair{"李松洪", "nr"}},
[]Pair{Pair{"一次性", "d"}, Pair{"交", "v"}, Pair{"多少", "m"}, Pair{"钱", "n"}},
[]Pair{Pair{"两块", "m"}, Pair{"五", "m"}, Pair{"一套", "m"}, Pair{"，", "x"}, Pair{"三块", "m"}, Pair{"八", "m"}, Pair{"一斤", "m"}, Pair{"，", "x"}, Pair{"四块", "m"}, Pair{"七", "m"}, Pair{"一本", "m"}, Pair{"，", "x"}, Pair{"五块", "m"}, Pair{"六", "m"}, Pair{"一条", "m"}},
[]Pair{Pair{"小", "a"}, Pair{"和尚", "nr"}, Pair{"留", "v"}, Pair{"了", "ul"}, Pair{"一个", "m"}, Pair{"像", "v"}, Pair{"大", "a"}, Pair{"和尚", "nr"}, Pair{"一样", "r"}, Pair{"的", "uj"}, Pair{"和尚头", "nr"}},
[]Pair{Pair{"我", "r"}, Pair{"是", "v"}, Pair{"中华人民共和国", "ns"}, Pair{"公民", "n"}, Pair{";", "x"}, Pair{"我", "r"}, Pair{"爸爸", "n"}, Pair{"是", "v"}, Pair{"共和党", "nt"}, Pair{"党员", "n"}, Pair{";", "x"}, Pair{" ", "x"}, Pair{"地铁", "n"}, Pair{"和平门", "ns"}, Pair{"站", "v"}},
[]Pair{Pair{"张晓梅", "nr"}, Pair{"去", "v"}, Pair{"人民", "n"}, Pair{"医院", "n"}, Pair{"做", "v"}, Pair{"了", "ul"}, Pair{"个", "q"}, Pair{"B超", "n"}, Pair{"然后", "c"}, Pair{"去", "v"}, Pair{"买", "v"}, Pair{"了", "ul"}, Pair{"件", "q"}, Pair{"T恤", "n"}},
[]Pair{Pair{"AT&T", "nz"}, Pair{"是", "v"}, Pair{"一件", "m"}, Pair{"不错", "a"}, Pair{"的", "uj"}, Pair{"公司", "n"}, Pair{"，", "x"}, Pair{"给", "p"}, Pair{"你", "r"}, Pair{"发", "v"}, Pair{"offer", "eng"}, Pair{"了", "ul"}, Pair{"吗", "y"}, Pair{"？", "x"}},
[]Pair{Pair{"C++", "nz"}, Pair{"和", "c"}, Pair{"c#", "nz"}, Pair{"是", "v"}, Pair{"什么", "r"}, Pair{"关系", "n"}, Pair{"？", "x"}, Pair{"11", "m"}, Pair{"+", "x"}, Pair{"122", "m"}, Pair{"=", "x"}, Pair{"133", "m"}, Pair{"，", "x"}, Pair{"是", "v"}, Pair{"吗", "y"}, Pair{"？", "x"}, Pair{"PI", "eng"}, Pair{"=", "x"}, Pair{"3.14159", "m"}},
[]Pair{Pair{"你", "r"}, Pair{"认识", "v"}, Pair{"那个", "r"}, Pair{"和", "c"}, Pair{"主席", "n"}, Pair{"握手", "v"}, Pair{"的", "uj"}, Pair{"的哥", "n"}, Pair{"吗", "y"}, Pair{"？", "x"}, Pair{"他", "r"}, Pair{"开", "v"}, Pair{"一辆", "m"}, Pair{"黑色", "n"}, Pair{"的士", "n"}, Pair{"。", "x"}},
[]Pair{Pair{"枪杆子", "n"}, Pair{"中", "f"}, Pair{"出", "v"}, Pair{"政权", "n"}},
}
noHMMCutResult = [][]WordTag{
[]WordTag{WordTag{"这", "r"}, WordTag{"是", "v"}, WordTag{"一个", "m"}, WordTag{"伸手不见五指", "i"}, WordTag{"的", "uj"}, WordTag{"黑夜", "n"}, WordTag{"。", "x"}, WordTag{"我", "r"}, WordTag{"叫", "v"}, WordTag{"孙悟空", "nr"}, WordTag{"，", "x"}, WordTag{"我", "r"}, WordTag{"爱", "v"}, WordTag{"北京", "ns"}, WordTag{"，", "x"}, WordTag{"我", "r"}, WordTag{"爱", "v"}, WordTag{"Python", "eng"}, WordTag{"和", "c"}, WordTag{"C++", "nz"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"不", "d"}, WordTag{"喜欢", "v"}, WordTag{"日本", "ns"}, WordTag{"和服", "nz"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"雷猴", "n"}, WordTag{"回归", "v"}, WordTag{"人间", "n"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"工信处", "n"}, WordTag{"女干事", "n"}, WordTag{"每月", "r"}, WordTag{"经过", "p"}, WordTag{"下属", "v"}, WordTag{"科室", "n"}, WordTag{"都", "d"}, WordTag{"要", "v"}, WordTag{"亲口", "n"}, WordTag{"交代", "n"}, WordTag{"24", "eng"}, WordTag{"口", "q"}, WordTag{"交换机", "n"}, WordTag{"等", "u"}, WordTag{"技术性", "n"}, WordTag{"器件", "n"}, WordTag{"的", "uj"}, WordTag{"安装", "v"}, WordTag{"工作", "vn"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"需要", "v"}, WordTag{"廉租房", "n"}},
[]WordTag{WordTag{"永和", "nz"}, WordTag{"服装", "vn"}, WordTag{"饰品", "n"}, WordTag{"有限公司", "n"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"爱", "v"}, WordTag{"北京", "ns"}, WordTag{"天安门", "ns"}},
[]WordTag{WordTag{"abc", "eng"}},
[]WordTag{WordTag{"隐", "n"}, WordTag{"马尔可夫", "nr"}},
[]WordTag{WordTag{"雷猴", "n"}, WordTag{"是", "v"}, WordTag{"个", "q"}, WordTag{"好", "a"}, WordTag{"网站", "n"}},
[]WordTag{WordTag{"“", "x"}, WordTag{"Microsoft", "eng"}, WordTag{"”", "x"}, WordTag{"一", "m"}, WordTag{"词", "n"}, WordTag{"由", "p"}, WordTag{"“", "x"}, WordTag{"MICROcomputer", "eng"}, WordTag{"（", "x"}, WordTag{"微型", "b"}, WordTag{"计算机", "n"}, WordTag{"）", "x"}, WordTag{"”", "x"}, WordTag{"和", "c"}, WordTag{"“", "x"}, WordTag{"SOFTware", "eng"}, WordTag{"（", "x"}, WordTag{"软件", "n"}, WordTag{"）", "x"}, WordTag{"”", "x"}, WordTag{"两", "m"}, WordTag{"部分", "n"}, WordTag{"组成", "v"}},
[]WordTag{WordTag{"草泥马", "n"}, WordTag{"和", "c"}, WordTag{"欺", "vn"}, WordTag{"实", "n"}, WordTag{"马", "n"}, WordTag{"是", "v"}, WordTag{"今年", "t"}, WordTag{"的", "uj"}, WordTag{"流行", "v"}, WordTag{"词汇", "n"}},
[]WordTag{WordTag{"伊", "ns"}, WordTag{"藤", "nr"}, WordTag{"洋华堂", "n"}, WordTag{"总府", "n"}, WordTag{"店", "n"}},
[]WordTag{WordTag{"中国科学院计算技术研究所", "nt"}},
[]WordTag{WordTag{"罗密欧", "nr"}, WordTag{"与", "p"}, WordTag{"朱丽叶", "nr"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"购买", "v"}, WordTag{"了", "ul"}, WordTag{"道具", "n"}, WordTag{"和", "c"}, WordTag{"服装", "vn"}},
[]WordTag{WordTag{"PS", "eng"}, WordTag{":", "x"}, WordTag{" ", "x"}, WordTag{"我", "r"}, WordTag{"觉得", "v"}, WordTag{"开源", "n"}, WordTag{"有", "v"}, WordTag{"一个", "m"}, WordTag{"好处", "d"}, WordTag{"，", "x"}, WordTag{"就是", "d"}, WordTag{"能够", "v"}, WordTag{"敦促", "v"}, WordTag{"自己", "r"}, WordTag{"不断改进", "l"}, WordTag{"，", "x"}, WordTag{"避免", "v"}, WordTag{"敞", "v"}, WordTag{"帚", "ng"}, WordTag{"自珍", "b"}},
[]WordTag{WordTag{"湖北省", "ns"}, WordTag{"石首市", "ns"}},
[]WordTag{WordTag{"湖北省", "ns"}, WordTag{"十堰市", "ns"}},
[]WordTag{WordTag{"总经理", "n"}, WordTag{"完成", "v"}, WordTag{"了", "ul"}, WordTag{"这件", "mq"}, WordTag{"事情", "n"}},
[]WordTag{WordTag{"电脑", "n"}, WordTag{"修好", "v"}, WordTag{"了", "ul"}},
[]WordTag{WordTag{"做好", "v"}, WordTag{"了", "ul"}, WordTag{"这件", "mq"}, WordTag{"事情", "n"}, WordTag{"就", "d"}, WordTag{"一了百了", "l"}, WordTag{"了", "ul"}},
[]WordTag{WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"我们", "r"}, WordTag{"买", "v"}, WordTag{"了", "ul"}, WordTag{"一个", "m"}, WordTag{"美的", "nr"}, WordTag{"空调", "n"}},
[]WordTag{WordTag{"线程", "n"}, WordTag{"初始化", "l"}, WordTag{"时", "n"}, WordTag{"我们", "r"}, WordTag{"要", "v"}, WordTag{"注意", "v"}},
[]WordTag{WordTag{"一个", "m"}, WordTag{"分子", "n"}, WordTag{"是", "v"}, WordTag{"由", "p"}, WordTag{"好多", "m"}, WordTag{"原子", "n"}, WordTag{"组织", "v"}, WordTag{"成", "n"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"祝", "v"}, WordTag{"你", "r"}, WordTag{"马到功成", "i"}},
[]WordTag{WordTag{"他", "r"}, WordTag{"掉", "zg"}, WordTag{"进", "v"}, WordTag{"了", "ul"}, WordTag{"无底洞", "ns"}, WordTag{"里", "f"}},
[]WordTag{WordTag{"中国", "ns"}, WordTag{"的", "uj"}, WordTag{"首都", "d"}, WordTag{"是", "v"}, WordTag{"北京", "ns"}},
[]WordTag{WordTag{"孙", "zg"}, WordTag{"君", "nz"}, WordTag{"意", "n"}},
[]WordTag{WordTag{"外交部", "nt"}, WordTag{"发言人", "l"}, WordTag{"马朝旭", "nr"}},
[]WordTag{WordTag{"领导人", "n"}, WordTag{"会议", "n"}, WordTag{"和", "c"}, WordTag{"第四届", "m"}, WordTag{"东亚", "ns"}, WordTag{"峰会", "n"}},
[]WordTag{WordTag{"在", "p"}, WordTag{"过去", "t"}, WordTag{"的", "uj"}, WordTag{"这", "r"}, WordTag{"五年", "t"}},
[]WordTag{WordTag{"还", "d"}, WordTag{"需要", "v"}, WordTag{"很", "zg"}, WordTag{"长", "a"}, WordTag{"的", "uj"}, WordTag{"路", "n"}, WordTag{"要", "v"}, WordTag{"走", "v"}},
[]WordTag{WordTag{"60", "eng"}, WordTag{"周年", "t"}, WordTag{"首都", "d"}, WordTag{"阅兵", "v"}},
[]WordTag{WordTag{"你好", "l"}, WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"买", "v"}, WordTag{"水果", "n"}, WordTag{"然后", "c"}, WordTag{"来", "v"}, WordTag{"世博园", "nr"}},
[]WordTag{WordTag{"买", "v"}, WordTag{"水果", "n"}, WordTag{"然后", "c"}, WordTag{"去", "v"}, WordTag{"世博园", "nr"}},
[]WordTag{WordTag{"但是", "c"}, WordTag{"后来", "t"}, WordTag{"我", "r"}, WordTag{"才", "d"}, WordTag{"知道", "v"}, WordTag{"你", "r"}, WordTag{"是", "v"}, WordTag{"对", "p"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"存在", "v"}, WordTag{"即", "v"}, WordTag{"合理", "vn"}},
[]WordTag{WordTag{"的", "uj"}, WordTag{"的", "uj"}, WordTag{"的", "uj"}, WordTag{"的", "uj"}, WordTag{"的", "uj"}, WordTag{"在", "p"}, WordTag{"的", "uj"}, WordTag{"的", "uj"}, WordTag{"的", "uj"}, WordTag{"的", "uj"}, WordTag{"就", "d"}, WordTag{"以", "p"}, WordTag{"和", "c"}, WordTag{"和", "c"}, WordTag{"和", "c"}},
[]WordTag{WordTag{"I", "eng"}, WordTag{" ", "x"}, WordTag{"love", "eng"}, WordTag{"你", "r"}, WordTag{"，", "x"}, WordTag{"不以为耻", "i"}, WordTag{"，", "x"}, WordTag{"反", "zg"}, WordTag{"以为", "c"}, WordTag{"rong", "eng"}},
[]WordTag{WordTag{"因", "p"}},
[]WordTag{},
[]WordTag{WordTag{"hello", "eng"}, WordTag{"你好", "l"}, WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"很", "zg"}, WordTag{"好", "a"}, WordTag{"但", "c"}, WordTag{"主要", "b"}, WordTag{"是", "v"}, WordTag{"基于", "p"}, WordTag{"网页", "n"}, WordTag{"形式", "n"}},
[]WordTag{WordTag{"hello", "eng"}, WordTag{"你好", "l"}, WordTag{"人们", "n"}, WordTag{"审美", "vn"}, WordTag{"的", "uj"}, WordTag{"观点", "n"}, WordTag{"是", "v"}, WordTag{"不同", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"为什么", "r"}, WordTag{"我", "r"}, WordTag{"不能", "v"}, WordTag{"拥有", "v"}, WordTag{"想要", "v"}, WordTag{"的", "uj"}, WordTag{"生活", "vn"}},
[]WordTag{WordTag{"后来", "t"}, WordTag{"我", "r"}, WordTag{"才", "d"}},
[]WordTag{WordTag{"此次", "r"}, WordTag{"来", "v"}, WordTag{"中国", "ns"}, WordTag{"是", "v"}, WordTag{"为了", "p"}},
[]WordTag{WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{",", "x"}, WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{"其实", "d"}, WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{"好人", "n"}, WordTag{"使用", "v"}, WordTag{"了", "ul"}, WordTag{"它", "r"}, WordTag{"就", "d"}, WordTag{"可以", "c"}, WordTag{"解决", "v"}, WordTag{"一些", "m"}, WordTag{"问题", "n"}},
[]WordTag{WordTag{"是因为", "c"}, WordTag{"和", "c"}, WordTag{"国家", "n"}},
[]WordTag{WordTag{"老年", "t"}, WordTag{"搜索", "v"}, WordTag{"还", "d"}, WordTag{"支持", "v"}},
[]WordTag{WordTag{"干脆", "d"}, WordTag{"就", "d"}, WordTag{"把", "p"}, WordTag{"那", "r"}, WordTag{"部", "n"}, WordTag{"蒙", "v"}, WordTag{"人", "n"}, WordTag{"的", "uj"}, WordTag{"闲", "n"}, WordTag{"法", "j"}, WordTag{"给", "p"}, WordTag{"废", "v"}, WordTag{"了", "ul"}, WordTag{"拉倒", "v"}, WordTag{"！", "x"}, WordTag{"RT", "eng"}, WordTag{" ", "x"}, WordTag{"@", "x"}, WordTag{"laoshipukong", "eng"}, WordTag{" ", "x"}, WordTag{":", "x"}, WordTag{" ", "x"}, WordTag{"27", "eng"}, WordTag{"日", "m"}, WordTag{"，", "x"}, WordTag{"全国人大常委会", "nt"}, WordTag{"第三次", "m"}, WordTag{"审议", "v"}, WordTag{"侵权", "v"}, WordTag{"责任法", "n"}, WordTag{"草案", "n"}, WordTag{"，", "x"}, WordTag{"删除", "v"}, WordTag{"了", "ul"}, WordTag{"有关", "vn"}, WordTag{"医疗", "n"}, WordTag{"损害", "v"}, WordTag{"责任", "n"}, WordTag{"“", "x"}, WordTag{"举证", "v"}, WordTag{"倒置", "v"}, WordTag{"”", "x"}, WordTag{"的", "uj"}, WordTag{"规定", "n"}, WordTag{"。", "x"}, WordTag{"在", "p"}, WordTag{"医患", "n"}, WordTag{"纠纷", "n"}, WordTag{"中", "f"}, WordTag{"本", "r"}, WordTag{"已", "d"}, WordTag{"处于", "v"}, WordTag{"弱势", "n"}, WordTag{"地位", "n"}, WordTag{"的", "uj"}, WordTag{"消费者", "n"}, WordTag{"由此", "c"}, WordTag{"将", "d"}, WordTag{"陷入", "v"}, WordTag{"万劫不复", "i"}, WordTag{"的", "uj"}, WordTag{"境地", "s"}, WordTag{"。", "x"}, WordTag{" ", "x"}},
[]WordTag{WordTag{"大", "a"}},
[]WordTag{},
[]WordTag{WordTag{"他", "r"}, WordTag{"说", "v"}, WordTag{"的", "uj"}, WordTag{"确实", "ad"}, WordTag{"在", "p"}, WordTag{"理", "n"}},
[]WordTag{WordTag{"长春", "ns"}, WordTag{"市长", "n"}, WordTag{"春节", "t"}, WordTag{"讲话", "n"}},
[]WordTag{WordTag{"结婚", "v"}, WordTag{"的", "uj"}, WordTag{"和", "c"}, WordTag{"尚未", "d"}, WordTag{"结婚", "v"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"结合", "v"}, WordTag{"成", "n"}, WordTag{"分子", "n"}, WordTag{"时", "n"}},
[]WordTag{WordTag{"旅游", "vn"}, WordTag{"和", "c"}, WordTag{"服务", "vn"}, WordTag{"是", "v"}, WordTag{"最好", "a"}, WordTag{"的", "uj"}},
[]WordTag{WordTag{"这件", "mq"}, WordTag{"事情", "n"}, WordTag{"的确", "d"}, WordTag{"是", "v"}, WordTag{"我", "r"}, WordTag{"的", "uj"}, WordTag{"错", "v"}},
[]WordTag{WordTag{"供", "v"}, WordTag{"大家", "n"}, WordTag{"参考", "v"}, WordTag{"指正", "v"}},
[]WordTag{WordTag{"哈尔滨", "ns"}, WordTag{"政府", "n"}, WordTag{"公布", "v"}, WordTag{"塌", "v"}, WordTag{"桥", "n"}, WordTag{"原因", "n"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"在", "p"}, WordTag{"机场", "n"}, WordTag{"入口处", "i"}},
[]WordTag{WordTag{"邢", "nr"}, WordTag{"永", "ns"}, WordTag{"臣", "n"}, WordTag{"摄影", "n"}, WordTag{"报道", "v"}},
[]WordTag{WordTag{"BP", "eng"}, WordTag{"神经网络", "n"}, WordTag{"如何", "r"}, WordTag{"训练", "vn"}, WordTag{"才能", "v"}, WordTag{"在", "p"}, WordTag{"分类", "n"}, WordTag{"时", "n"}, WordTag{"增加", "v"}, WordTag{"区分度", "n"}, WordTag{"？", "x"}},
[]WordTag{WordTag{"南京市", "ns"}, WordTag{"长江大桥", "ns"}},
[]WordTag{WordTag{"应", "v"}, WordTag{"一些", "m"}, WordTag{"使用者", "n"}, WordTag{"的", "uj"}, WordTag{"建议", "n"}, WordTag{"，", "x"}, WordTag{"也", "d"}, WordTag{"为了", "p"}, WordTag{"便于", "v"}, WordTag{"利用", "n"}, WordTag{"NiuTrans", "eng"}, WordTag{"用于", "v"}, WordTag{"SMT", "eng"}, WordTag{"研究", "vn"}},
[]WordTag{WordTag{"长春市", "ns"}, WordTag{"长春", "ns"}, WordTag{"药店", "n"}},
[]WordTag{WordTag{"邓颖超", "nr"}, WordTag{"生前", "t"}, WordTag{"最", "d"}, WordTag{"喜欢", "v"}, WordTag{"的", "uj"}, WordTag{"衣服", "n"}},
[]WordTag{WordTag{"胡锦涛", "nr"}, WordTag{"是", "v"}, WordTag{"热爱", "a"}, WordTag{"世界", "n"}, WordTag{"和平", "nz"}, WordTag{"的", "uj"}, WordTag{"政治局", "n"}, WordTag{"常委", "j"}},
[]WordTag{WordTag{"程序员", "n"}, WordTag{"祝", "v"}, WordTag{"海林", "nz"}, WordTag{"和", "c"}, WordTag{"朱", "nr"}, WordTag{"会", "v"}, WordTag{"震", "v"}, WordTag{"是", "v"}, WordTag{"在", "p"}, WordTag{"孙", "zg"}, WordTag{"健", "a"}, WordTag{"的", "uj"}, WordTag{"左面", "f"}, WordTag{"和", "c"}, WordTag{"右面", "f"}, WordTag{",", "x"}, WordTag{" ", "x"}, WordTag{"范", "nr"}, WordTag{"凯", "nr"}, WordTag{"在", "p"}, WordTag{"最", "d"}, WordTag{"右面", "f"}, WordTag{".", "x"}, WordTag{"再", "d"}, WordTag{"往", "zg"}, WordTag{"左", "m"}, WordTag{"是", "v"}, WordTag{"李", "nr"}, WordTag{"松", "v"}, WordTag{"洪", "nr"}},
[]WordTag{WordTag{"一次性", "d"}, WordTag{"交", "v"}, WordTag{"多少", "m"}, WordTag{"钱", "n"}},
[]WordTag{WordTag{"两块", "m"}, WordTag{"五", "m"}, WordTag{"一套", "m"}, WordTag{"，", "x"}, WordTag{"三块", "m"}, WordTag{"八", "m"}, WordTag{"一斤", "m"}, WordTag{"，", "x"}, WordTag{"四块", "m"}, WordTag{"七", "m"}, WordTag{"一本", "m"}, WordTag{"，", "x"}, WordTag{"五块", "m"}, WordTag{"六", "m"}, WordTag{"一条", "m"}},
[]WordTag{WordTag{"小", "a"}, WordTag{"和尚", "nr"}, WordTag{"留", "v"}, WordTag{"了", "ul"}, WordTag{"一个", "m"}, WordTag{"像", "v"}, WordTag{"大", "a"}, WordTag{"和尚", "nr"}, WordTag{"一样", "r"}, WordTag{"的", "uj"}, WordTag{"和尚头", "nr"}},
[]WordTag{WordTag{"我", "r"}, WordTag{"是", "v"}, WordTag{"中华人民共和国", "ns"}, WordTag{"公民", "n"}, WordTag{";", "x"}, WordTag{"我", "r"}, WordTag{"爸爸", "n"}, WordTag{"是", "v"}, WordTag{"共和党", "nt"}, WordTag{"党员", "n"}, WordTag{";", "x"}, WordTag{" ", "x"}, WordTag{"地铁", "n"}, WordTag{"和平门", "ns"}, WordTag{"站", "v"}},
[]WordTag{WordTag{"张晓梅", "nr"}, WordTag{"去", "v"}, WordTag{"人民", "n"}, WordTag{"医院", "n"}, WordTag{"做", "v"}, WordTag{"了", "ul"}, WordTag{"个", "q"}, WordTag{"B超", "n"}, WordTag{"然后", "c"}, WordTag{"去", "v"}, WordTag{"买", "v"}, WordTag{"了", "ul"}, WordTag{"件", "zg"}, WordTag{"T恤", "n"}},
[]WordTag{WordTag{"AT&T", "nz"}, WordTag{"是", "v"}, WordTag{"一件", "m"}, WordTag{"不错", "a"}, WordTag{"的", "uj"}, WordTag{"公司", "n"}, WordTag{"，", "x"}, WordTag{"给", "p"}, WordTag{"你", "r"}, WordTag{"发", "v"}, WordTag{"offer", "eng"}, WordTag{"了", "ul"}, WordTag{"吗", "y"}, WordTag{"？", "x"}},
[]WordTag{WordTag{"C++", "nz"}, WordTag{"和", "c"}, WordTag{"c#", "nz"}, WordTag{"是", "v"}, WordTag{"什么", "r"}, WordTag{"关系", "n"}, WordTag{"？", "x"}, WordTag{"11", "eng"}, WordTag{"+", "x"}, WordTag{"122", "eng"}, WordTag{"=", "x"}, WordTag{"133", "eng"}, WordTag{"，", "x"}, WordTag{"是", "v"}, WordTag{"吗", "y"}, WordTag{"？", "x"}, WordTag{"PI", "eng"}, WordTag{"=", "x"}, WordTag{"3", "eng"}, WordTag{".", "x"}, WordTag{"14159", "eng"}},
[]WordTag{WordTag{"你", "r"}, WordTag{"认识", "v"}, WordTag{"那个", "r"}, WordTag{"和", "c"}, WordTag{"主席", "n"}, WordTag{"握手", "v"}, WordTag{"的", "uj"}, WordTag{"的哥", "n"}, WordTag{"吗", "y"}, WordTag{"？", "x"}, WordTag{"他", "r"}, WordTag{"开", "v"}, WordTag{"一辆", "m"}, WordTag{"黑色", "n"}, WordTag{"的士", "n"}, WordTag{"。", "x"}},
[]WordTag{WordTag{"枪杆子", "n"}, WordTag{"中", "f"}, WordTag{"出", "v"}, WordTag{"政权", "n"}},
noHMMCutResult = [][]Pair{
[]Pair{Pair{"这", "r"}, Pair{"是", "v"}, Pair{"一个", "m"}, Pair{"伸手不见五指", "i"}, Pair{"的", "uj"}, Pair{"黑夜", "n"}, Pair{"。", "x"}, Pair{"我", "r"}, Pair{"叫", "v"}, Pair{"孙悟空", "nr"}, Pair{"，", "x"}, Pair{"我", "r"}, Pair{"爱", "v"}, Pair{"北京", "ns"}, Pair{"，", "x"}, Pair{"我", "r"}, Pair{"爱", "v"}, Pair{"Python", "eng"}, Pair{"和", "c"}, Pair{"C++", "nz"}, Pair{"。", "x"}},
[]Pair{Pair{"我", "r"}, Pair{"不", "d"}, Pair{"喜欢", "v"}, Pair{"日本", "ns"}, Pair{"和服", "nz"}, Pair{"。", "x"}},
[]Pair{Pair{"雷猴", "n"}, Pair{"回归", "v"}, Pair{"人间", "n"}, Pair{"。", "x"}},
[]Pair{Pair{"工信处", "n"}, Pair{"女干事", "n"}, Pair{"每月", "r"}, Pair{"经过", "p"}, Pair{"下属", "v"}, Pair{"科室", "n"}, Pair{"都", "d"}, Pair{"要", "v"}, Pair{"亲口", "n"}, Pair{"交代", "n"}, Pair{"24", "eng"}, Pair{"口", "q"}, Pair{"交换机", "n"}, Pair{"等", "u"}, Pair{"技术性", "n"}, Pair{"器件", "n"}, Pair{"的", "uj"}, Pair{"安装", "v"}, Pair{"工作", "vn"}},
[]Pair{Pair{"我", "r"}, Pair{"需要", "v"}, Pair{"廉租房", "n"}},
[]Pair{Pair{"永和", "nz"}, Pair{"服装", "vn"}, Pair{"饰品", "n"}, Pair{"有限公司", "n"}},
[]Pair{Pair{"我", "r"}, Pair{"爱", "v"}, Pair{"北京", "ns"}, Pair{"天安门", "ns"}},
[]Pair{Pair{"abc", "eng"}},
[]Pair{Pair{"隐", "n"}, Pair{"马尔可夫", "nr"}},
[]Pair{Pair{"雷猴", "n"}, Pair{"是", "v"}, Pair{"个", "q"}, Pair{"好", "a"}, Pair{"网站", "n"}},
[]Pair{Pair{"“", "x"}, Pair{"Microsoft", "eng"}, Pair{"”", "x"}, Pair{"一", "m"}, Pair{"词", "n"}, Pair{"由", "p"}, Pair{"“", "x"}, Pair{"MICROcomputer", "eng"}, Pair{"（", "x"}, Pair{"微型", "b"}, Pair{"计算机", "n"}, Pair{"）", "x"}, Pair{"”", "x"}, Pair{"和", "c"}, Pair{"“", "x"}, Pair{"SOFTware", "eng"}, Pair{"（", "x"}, Pair{"软件", "n"}, Pair{"）", "x"}, Pair{"”", "x"}, Pair{"两", "m"}, Pair{"部分", "n"}, Pair{"组成", "v"}},
[]Pair{Pair{"草泥马", "n"}, Pair{"和", "c"}, Pair{"欺", "vn"}, Pair{"实", "n"}, Pair{"马", "n"}, Pair{"是", "v"}, Pair{"今年", "t"}, Pair{"的", "uj"}, Pair{"流行", "v"}, Pair{"词汇", "n"}},
[]Pair{Pair{"伊", "ns"}, Pair{"藤", "nr"}, Pair{"洋华堂", "n"}, Pair{"总府", "n"}, Pair{"店", "n"}},
[]Pair{Pair{"中国科学院计算技术研究所", "nt"}},
[]Pair{Pair{"罗密欧", "nr"}, Pair{"与", "p"}, Pair{"朱丽叶", "nr"}},
[]Pair{Pair{"我", "r"}, Pair{"购买", "v"}, Pair{"了", "ul"}, Pair{"道具", "n"}, Pair{"和", "c"}, Pair{"服装", "vn"}},
[]Pair{Pair{"PS", "eng"}, Pair{":", "x"}, Pair{" ", "x"}, Pair{"我", "r"}, Pair{"觉得", "v"}, Pair{"开源", "n"}, Pair{"有", "v"}, Pair{"一个", "m"}, Pair{"好处", "d"}, Pair{"，", "x"}, Pair{"就是", "d"}, Pair{"能够", "v"}, Pair{"敦促", "v"}, Pair{"自己", "r"}, Pair{"不断改进", "l"}, Pair{"，", "x"}, Pair{"避免", "v"}, Pair{"敞", "v"}, Pair{"帚", "ng"}, Pair{"自珍", "b"}},
[]Pair{Pair{"湖北省", "ns"}, Pair{"石首市", "ns"}},
[]Pair{Pair{"湖北省", "ns"}, Pair{"十堰市", "ns"}},
[]Pair{Pair{"总经理", "n"}, Pair{"完成", "v"}, Pair{"了", "ul"}, Pair{"这件", "mq"}, Pair{"事情", "n"}},
[]Pair{Pair{"电脑", "n"}, Pair{"修好", "v"}, Pair{"了", "ul"}},
[]Pair{Pair{"做好", "v"}, Pair{"了", "ul"}, Pair{"这件", "mq"}, Pair{"事情", "n"}, Pair{"就", "d"}, Pair{"一了百了", "l"}, Pair{"了", "ul"}},
[]Pair{Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"我们", "r"}, Pair{"买", "v"}, Pair{"了", "ul"}, Pair{"一个", "m"}, Pair{"美的", "nr"}, Pair{"空调", "n"}},
[]Pair{Pair{"线程", "n"}, Pair{"初始化", "l"}, Pair{"时", "n"}, Pair{"我们", "r"}, Pair{"要", "v"}, Pair{"注意", "v"}},
[]Pair{Pair{"一个", "m"}, Pair{"分子", "n"}, Pair{"是", "v"}, Pair{"由", "p"}, Pair{"好多", "m"}, Pair{"原子", "n"}, Pair{"组织", "v"}, Pair{"成", "n"}, Pair{"的", "uj"}},
[]Pair{Pair{"祝", "v"}, Pair{"你", "r"}, Pair{"马到功成", "i"}},
[]Pair{Pair{"他", "r"}, Pair{"掉", "zg"}, Pair{"进", "v"}, Pair{"了", "ul"}, Pair{"无底洞", "ns"}, Pair{"里", "f"}},
[]Pair{Pair{"中国", "ns"}, Pair{"的", "uj"}, Pair{"首都", "d"}, Pair{"是", "v"}, Pair{"北京", "ns"}},
[]Pair{Pair{"孙", "zg"}, Pair{"君", "nz"}, Pair{"意", "n"}},
[]Pair{Pair{"外交部", "nt"}, Pair{"发言人", "l"}, Pair{"马朝旭", "nr"}},
[]Pair{Pair{"领导人", "n"}, Pair{"会议", "n"}, Pair{"和", "c"}, Pair{"第四届", "m"}, Pair{"东亚", "ns"}, Pair{"峰会", "n"}},
[]Pair{Pair{"在", "p"}, Pair{"过去", "t"}, Pair{"的", "uj"}, Pair{"这", "r"}, Pair{"五年", "t"}},
[]Pair{Pair{"还", "d"}, Pair{"需要", "v"}, Pair{"很", "zg"}, Pair{"长", "a"}, Pair{"的", "uj"}, Pair{"路", "n"}, Pair{"要", "v"}, Pair{"走", "v"}},
[]Pair{Pair{"60", "eng"}, Pair{"周年", "t"}, Pair{"首都", "d"}, Pair{"阅兵", "v"}},
[]Pair{Pair{"你好", "l"}, Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"买", "v"}, Pair{"水果", "n"}, Pair{"然后", "c"}, Pair{"来", "v"}, Pair{"世博园", "nr"}},
[]Pair{Pair{"买", "v"}, Pair{"水果", "n"}, Pair{"然后", "c"}, Pair{"去", "v"}, Pair{"世博园", "nr"}},
[]Pair{Pair{"但是", "c"}, Pair{"后来", "t"}, Pair{"我", "r"}, Pair{"才", "d"}, Pair{"知道", "v"}, Pair{"你", "r"}, Pair{"是", "v"}, Pair{"对", "p"}, Pair{"的", "uj"}},
[]Pair{Pair{"存在", "v"}, Pair{"即", "v"}, Pair{"合理", "vn"}},
[]Pair{Pair{"的", "uj"}, Pair{"的", "uj"}, Pair{"的", "uj"}, Pair{"的", "uj"}, Pair{"的", "uj"}, Pair{"在", "p"}, Pair{"的", "uj"}, Pair{"的", "uj"}, Pair{"的", "uj"}, Pair{"的", "uj"}, Pair{"就", "d"}, Pair{"以", "p"}, Pair{"和", "c"}, Pair{"和", "c"}, Pair{"和", "c"}},
[]Pair{Pair{"I", "eng"}, Pair{" ", "x"}, Pair{"love", "eng"}, Pair{"你", "r"}, Pair{"，", "x"}, Pair{"不以为耻", "i"}, Pair{"，", "x"}, Pair{"反", "zg"}, Pair{"以为", "c"}, Pair{"rong", "eng"}},
[]Pair{Pair{"因", "p"}},
[]Pair{},
[]Pair{Pair{"hello", "eng"}, Pair{"你好", "l"}, Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"很", "zg"}, Pair{"好", "a"}, Pair{"但", "c"}, Pair{"主要", "b"}, Pair{"是", "v"}, Pair{"基于", "p"}, Pair{"网页", "n"}, Pair{"形式", "n"}},
[]Pair{Pair{"hello", "eng"}, Pair{"你好", "l"}, Pair{"人们", "n"}, Pair{"审美", "vn"}, Pair{"的", "uj"}, Pair{"观点", "n"}, Pair{"是", "v"}, Pair{"不同", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"为什么", "r"}, Pair{"我", "r"}, Pair{"不能", "v"}, Pair{"拥有", "v"}, Pair{"想要", "v"}, Pair{"的", "uj"}, Pair{"生活", "vn"}},
[]Pair{Pair{"后来", "t"}, Pair{"我", "r"}, Pair{"才", "d"}},
[]Pair{Pair{"此次", "r"}, Pair{"来", "v"}, Pair{"中国", "ns"}, Pair{"是", "v"}, Pair{"为了", "p"}},
[]Pair{Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{",", "x"}, Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{"其实", "d"}, Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{"好人", "n"}, Pair{"使用", "v"}, Pair{"了", "ul"}, Pair{"它", "r"}, Pair{"就", "d"}, Pair{"可以", "c"}, Pair{"解决", "v"}, Pair{"一些", "m"}, Pair{"问题", "n"}},
[]Pair{Pair{"是因为", "c"}, Pair{"和", "c"}, Pair{"国家", "n"}},
[]Pair{Pair{"老年", "t"}, Pair{"搜索", "v"}, Pair{"还", "d"}, Pair{"支持", "v"}},
[]Pair{Pair{"干脆", "d"}, Pair{"就", "d"}, Pair{"把", "p"}, Pair{"那", "r"}, Pair{"部", "n"}, Pair{"蒙", "v"}, Pair{"人", "n"}, Pair{"的", "uj"}, Pair{"闲", "n"}, Pair{"法", "j"}, Pair{"给", "p"}, Pair{"废", "v"}, Pair{"了", "ul"}, Pair{"拉倒", "v"}, Pair{"！", "x"}, Pair{"RT", "eng"}, Pair{" ", "x"}, Pair{"@", "x"}, Pair{"laoshipukong", "eng"}, Pair{" ", "x"}, Pair{":", "x"}, Pair{" ", "x"}, Pair{"27", "eng"}, Pair{"日", "m"}, Pair{"，", "x"}, Pair{"全国人大常委会", "nt"}, Pair{"第三次", "m"}, Pair{"审议", "v"}, Pair{"侵权", "v"}, Pair{"责任法", "n"}, Pair{"草案", "n"}, Pair{"，", "x"}, Pair{"删除", "v"}, Pair{"了", "ul"}, Pair{"有关", "vn"}, Pair{"医疗", "n"}, Pair{"损害", "v"}, Pair{"责任", "n"}, Pair{"“", "x"}, Pair{"举证", "v"}, Pair{"倒置", "v"}, Pair{"”", "x"}, Pair{"的", "uj"}, Pair{"规定", "n"}, Pair{"。", "x"}, Pair{"在", "p"}, Pair{"医患", "n"}, Pair{"纠纷", "n"}, Pair{"中", "f"}, Pair{"本", "r"}, Pair{"已", "d"}, Pair{"处于", "v"}, Pair{"弱势", "n"}, Pair{"地位", "n"}, Pair{"的", "uj"}, Pair{"消费者", "n"}, Pair{"由此", "c"}, Pair{"将", "d"}, Pair{"陷入", "v"}, Pair{"万劫不复", "i"}, Pair{"的", "uj"}, Pair{"境地", "s"}, Pair{"。", "x"}, Pair{" ", "x"}},
[]Pair{Pair{"大", "a"}},
[]Pair{},
[]Pair{Pair{"他", "r"}, Pair{"说", "v"}, Pair{"的", "uj"}, Pair{"确实", "ad"}, Pair{"在", "p"}, Pair{"理", "n"}},
[]Pair{Pair{"长春", "ns"}, Pair{"市长", "n"}, Pair{"春节", "t"}, Pair{"讲话", "n"}},
[]Pair{Pair{"结婚", "v"}, Pair{"的", "uj"}, Pair{"和", "c"}, Pair{"尚未", "d"}, Pair{"结婚", "v"}, Pair{"的", "uj"}},
[]Pair{Pair{"结合", "v"}, Pair{"成", "n"}, Pair{"分子", "n"}, Pair{"时", "n"}},
[]Pair{Pair{"旅游", "vn"}, Pair{"和", "c"}, Pair{"服务", "vn"}, Pair{"是", "v"}, Pair{"最好", "a"}, Pair{"的", "uj"}},
[]Pair{Pair{"这件", "mq"}, Pair{"事情", "n"}, Pair{"的确", "d"}, Pair{"是", "v"}, Pair{"我", "r"}, Pair{"的", "uj"}, Pair{"错", "v"}},
[]Pair{Pair{"供", "v"}, Pair{"大家", "n"}, Pair{"参考", "v"}, Pair{"指正", "v"}},
[]Pair{Pair{"哈尔滨", "ns"}, Pair{"政府", "n"}, Pair{"公布", "v"}, Pair{"塌", "v"}, Pair{"桥", "n"}, Pair{"原因", "n"}},
[]Pair{Pair{"我", "r"}, Pair{"在", "p"}, Pair{"机场", "n"}, Pair{"入口处", "i"}},
[]Pair{Pair{"邢", "nr"}, Pair{"永", "ns"}, Pair{"臣", "n"}, Pair{"摄影", "n"}, Pair{"报道", "v"}},
[]Pair{Pair{"BP", "eng"}, Pair{"神经网络", "n"}, Pair{"如何", "r"}, Pair{"训练", "vn"}, Pair{"才能", "v"}, Pair{"在", "p"}, Pair{"分类", "n"}, Pair{"时", "n"}, Pair{"增加", "v"}, Pair{"区分度", "n"}, Pair{"？", "x"}},
[]Pair{Pair{"南京市", "ns"}, Pair{"长江大桥", "ns"}},
[]Pair{Pair{"应", "v"}, Pair{"一些", "m"}, Pair{"使用者", "n"}, Pair{"的", "uj"}, Pair{"建议", "n"}, Pair{"，", "x"}, Pair{"也", "d"}, Pair{"为了", "p"}, Pair{"便于", "v"}, Pair{"利用", "n"}, Pair{"NiuTrans", "eng"}, Pair{"用于", "v"}, Pair{"SMT", "eng"}, Pair{"研究", "vn"}},
[]Pair{Pair{"长春市", "ns"}, Pair{"长春", "ns"}, Pair{"药店", "n"}},
[]Pair{Pair{"邓颖超", "nr"}, Pair{"生前", "t"}, Pair{"最", "d"}, Pair{"喜欢", "v"}, Pair{"的", "uj"}, Pair{"衣服", "n"}},
[]Pair{Pair{"胡锦涛", "nr"}, Pair{"是", "v"}, Pair{"热爱", "a"}, Pair{"世界", "n"}, Pair{"和平", "nz"}, Pair{"的", "uj"}, Pair{"政治局", "n"}, Pair{"常委", "j"}},
[]Pair{Pair{"程序员", "n"}, Pair{"祝", "v"}, Pair{"海林", "nz"}, Pair{"和", "c"}, Pair{"朱", "nr"}, Pair{"会", "v"}, Pair{"震", "v"}, Pair{"是", "v"}, Pair{"在", "p"}, Pair{"孙", "zg"}, Pair{"健", "a"}, Pair{"的", "uj"}, Pair{"左面", "f"}, Pair{"和", "c"}, Pair{"右面", "f"}, Pair{",", "x"}, Pair{" ", "x"}, Pair{"范", "nr"}, Pair{"凯", "nr"}, Pair{"在", "p"}, Pair{"最", "d"}, Pair{"右面", "f"}, Pair{".", "x"}, Pair{"再", "d"}, Pair{"往", "zg"}, Pair{"左", "m"}, Pair{"是", "v"}, Pair{"李", "nr"}, Pair{"松", "v"}, Pair{"洪", "nr"}},
[]Pair{Pair{"一次性", "d"}, Pair{"交", "v"}, Pair{"多少", "m"}, Pair{"钱", "n"}},
[]Pair{Pair{"两块", "m"}, Pair{"五", "m"}, Pair{"一套", "m"}, Pair{"，", "x"}, Pair{"三块", "m"}, Pair{"八", "m"}, Pair{"一斤", "m"}, Pair{"，", "x"}, Pair{"四块", "m"}, Pair{"七", "m"}, Pair{"一本", "m"}, Pair{"，", "x"}, Pair{"五块", "m"}, Pair{"六", "m"}, Pair{"一条", "m"}},
[]Pair{Pair{"小", "a"}, Pair{"和尚", "nr"}, Pair{"留", "v"}, Pair{"了", "ul"}, Pair{"一个", "m"}, Pair{"像", "v"}, Pair{"大", "a"}, Pair{"和尚", "nr"}, Pair{"一样", "r"}, Pair{"的", "uj"}, Pair{"和尚头", "nr"}},
[]Pair{Pair{"我", "r"}, Pair{"是", "v"}, Pair{"中华人民共和国", "ns"}, Pair{"公民", "n"}, Pair{";", "x"}, Pair{"我", "r"}, Pair{"爸爸", "n"}, Pair{"是", "v"}, Pair{"共和党", "nt"}, Pair{"党员", "n"}, Pair{";", "x"}, Pair{" ", "x"}, Pair{"地铁", "n"}, Pair{"和平门", "ns"}, Pair{"站", "v"}},
[]Pair{Pair{"张晓梅", "nr"}, Pair{"去", "v"}, Pair{"人民", "n"}, Pair{"医院", "n"}, Pair{"做", "v"}, Pair{"了", "ul"}, Pair{"个", "q"}, Pair{"B超", "n"}, Pair{"然后", "c"}, Pair{"去", "v"}, Pair{"买", "v"}, Pair{"了", "ul"}, Pair{"件", "zg"}, Pair{"T恤", "n"}},
[]Pair{Pair{"AT&T", "nz"}, Pair{"是", "v"}, Pair{"一件", "m"}, Pair{"不错", "a"}, Pair{"的", "uj"}, Pair{"公司", "n"}, Pair{"，", "x"}, Pair{"给", "p"}, Pair{"你", "r"}, Pair{"发", "v"}, Pair{"offer", "eng"}, Pair{"了", "ul"}, Pair{"吗", "y"}, Pair{"？", "x"}},
[]Pair{Pair{"C++", "nz"}, Pair{"和", "c"}, Pair{"c#", "nz"}, Pair{"是", "v"}, Pair{"什么", "r"}, Pair{"关系", "n"}, Pair{"？", "x"}, Pair{"11", "eng"}, Pair{"+", "x"}, Pair{"122", "eng"}, Pair{"=", "x"}, Pair{"133", "eng"}, Pair{"，", "x"}, Pair{"是", "v"}, Pair{"吗", "y"}, Pair{"？", "x"}, Pair{"PI", "eng"}, Pair{"=", "x"}, Pair{"3", "eng"}, Pair{".", "x"}, Pair{"14159", "eng"}},
[]Pair{Pair{"你", "r"}, Pair{"认识", "v"}, Pair{"那个", "r"}, Pair{"和", "c"}, Pair{"主席", "n"}, Pair{"握手", "v"}, Pair{"的", "uj"}, Pair{"的哥", "n"}, Pair{"吗", "y"}, Pair{"？", "x"}, Pair{"他", "r"}, Pair{"开", "v"}, Pair{"一辆", "m"}, Pair{"黑色", "n"}, Pair{"的士", "n"}, Pair{"。", "x"}},
[]Pair{Pair{"枪杆子", "n"}, Pair{"中", "f"}, Pair{"出", "v"}, Pair{"政权", "n"}},
}
)
func chanToArray(ch chan WordTag) []WordTag {
result := make([]WordTag, 0)
func chanToArray(ch chan Pair) []Pair {
result := make([]Pair, 0)
for word := range ch {
result = append(result, word)
}
@@ -309,11 +309,11 @@ func TestBug132(t *testing.T) {
*/
p, _ := NewPosseg("../dict.txt")
sentence := "又跛又啞"
cutResult := []WordTag{
WordTag{"又", "d"},
WordTag{"跛", "a"},
WordTag{"又", "d"},
WordTag{"啞", "v"},
cutResult := []Pair{
Pair{"又", "d"},
Pair{"跛", "a"},
Pair{"又", "d"},
Pair{"啞", "v"},
}
result := chanToArray(p.Cut(sentence, true))
if len(cutResult) != len(result) {
@@ -332,20 +332,20 @@ func TestBug137(t *testing.T) {
*/
p, _ := NewPosseg("../dict.txt")
sentence := "前港督衛奕信在八八年十月宣布成立中央政策研究組"
cutResult := []WordTag{
WordTag{"前", "f"},
WordTag{"港督", "n"},
WordTag{"衛奕", "z"},
WordTag{"信", "n"},
WordTag{"在", "p"},
WordTag{"八八年", "m"},
WordTag{"十月", "t"},
WordTag{"宣布", "v"},
WordTag{"成立", "v"},
WordTag{"中央", "n"},
WordTag{"政策", "n"},
WordTag{"研究", "vn"},
WordTag{"組", "x"},
cutResult := []Pair{
Pair{"前", "f"},
Pair{"港督", "n"},
Pair{"衛奕", "z"},
Pair{"信", "n"},
Pair{"在", "p"},
Pair{"八八年", "m"},
Pair{"十月", "t"},
Pair{"宣布", "v"},
Pair{"成立", "v"},
Pair{"中央", "n"},
Pair{"政策", "n"},
Pair{"研究", "vn"},
Pair{"組", "x"},
}
result := chanToArray(p.Cut(sentence, true))
if len(cutResult) != len(result) {
@@ -363,44 +363,44 @@ func TestUserDict(t *testing.T) {
p.LoadUserDict("../userdict.txt")
sentence := "李小福是创新办主任也是云计算方面的专家; 什么是八一双鹿例如我输入一个带“韩玉赏鉴”的标题，在自定义词库中也增加了此词为N类型"
cutResult := []WordTag{
WordTag{"李小福", "nr"},
WordTag{"是", "v"},
WordTag{"创新办", "i"},
WordTag{"主任", "b"},
WordTag{"也", "d"},
WordTag{"是", "v"},
WordTag{"云计算", "x"},
WordTag{"方面", "n"},
WordTag{"的", "uj"},
WordTag{"专家", "n"},
WordTag{";", "x"},
WordTag{" ", "x"},
WordTag{"什么", "r"},
WordTag{"是", "v"},
WordTag{"八一双鹿", "nz"},
WordTag{"例如", "v"},
WordTag{"我", "r"},
WordTag{"输入", "v"},
WordTag{"一个", "m"},
WordTag{"带", "v"},
WordTag{"“", "x"},
WordTag{"韩玉赏鉴", "nz"},
WordTag{"”", "x"},
WordTag{"的", "uj"},
WordTag{"标题", "n"},
WordTag{"，", "x"},
WordTag{"在", "p"},
WordTag{"自定义词", "n"},
WordTag{"库中", "nrt"},
WordTag{"也", "d"},
WordTag{"增加", "v"},
WordTag{"了", "ul"},
WordTag{"此", "r"},
WordTag{"词", "n"},
WordTag{"为", "p"},
WordTag{"N", "eng"},
WordTag{"类型", "n"}}
cutResult := []Pair{
Pair{"李小福", "nr"},
Pair{"是", "v"},
Pair{"创新办", "i"},
Pair{"主任", "b"},
Pair{"也", "d"},
Pair{"是", "v"},
Pair{"云计算", "x"},
Pair{"方面", "n"},
Pair{"的", "uj"},
Pair{"专家", "n"},
Pair{";", "x"},
Pair{" ", "x"},
Pair{"什么", "r"},
Pair{"是", "v"},
Pair{"八一双鹿", "nz"},
Pair{"例如", "v"},
Pair{"我", "r"},
Pair{"输入", "v"},
Pair{"一个", "m"},
Pair{"带", "v"},
Pair{"“", "x"},
Pair{"韩玉赏鉴", "nz"},
Pair{"”", "x"},
Pair{"的", "uj"},
Pair{"标题", "n"},
Pair{"，", "x"},
Pair{"在", "p"},
Pair{"自定义词", "n"},
Pair{"库中", "nrt"},
Pair{"也", "d"},
Pair{"增加", "v"},
Pair{"了", "ul"},
Pair{"此", "r"},
Pair{"词", "n"},
Pair{"为", "p"},
Pair{"N", "eng"},
Pair{"类型", "n"}}
result := chanToArray(p.Cut(sentence, true))
if len(cutResult) != len(result) {

3
util.go
View File

@@ -40,13 +40,12 @@ func LoadDict(l DictLoader, dictFileName string, usingFlag bool) error {
}
defer dictFile.Close()
scanner := bufio.NewScanner(dictFile)
var entry *Entry
var entry Entry
var line string
var fields []string
for scanner.Scan() {
line = scanner.Text()
fields = strings.Split(line, " ")
entry = NewEntry()
entry.Word = strings.Replace(fields[0], "\ufeff", "", 1)
if length := len(fields); length > 1 {
entry.Freq, err = strconv.ParseFloat(fields[1], 64)