code refactor, added more documents

finalseg/finalseg.go

@@ -13,10 +13,10 @@ func cutHan(sentence string) chan string {
 	result := make(chan string)
 	go func() {
 		runes := []rune(sentence)
-		_, pos_list := viterbi(runes, []byte{'B', 'M', 'E', 'S'})
+		_, posList := viterbi(runes, []byte{'B', 'M', 'E', 'S'})
 		begin, next := 0, 0
 		for i, char := range runes {
-			pos := pos_list[i]
+			pos := posList[i]
 			switch pos {
 			case 'B':
 				begin = i
@@ -36,6 +36,8 @@ func cutHan(sentence string) chan string {
 	return result
 }
 
+// Cut cuts sentence into words using Hidden Markov Model with Viterbi
+// algorithm. It is used by Jiebago for unknonw words.
 func Cut(sentence string) chan string {
 	result := make(chan string)
 	s := sentence

finalseg/finalseg_test.go

@@ -6,7 +6,7 @@ import (
 )
 
 func chanToArray(ch chan string) []string {
-	result := make([]string, 0)
+	var result []string
 	for word := range ch {
 		result = append(result, word)
 	}

finalseg/viterbi.go

@@ -67,11 +67,11 @@ func viterbi(obs []rune, states []byte) (float64, []byte) {
 		V[t] = make(map[byte]float64)
 		for _, y := range states {
 			ps0 := make(probStates, 0)
-			var em_p float64
+			var emP float64
 			if val, ok := probEmit[y][obs[t]]; ok {
-				em_p = val
+				emP = val
 			} else {
-				em_p = minFloat
+				emP = minFloat
 			}
 			for _, y0 := range prevStatus[y] {
 				var transP float64
@@ -80,7 +80,7 @@ func viterbi(obs []rune, states []byte) (float64, []byte) {
 				} else {
 					transP = minFloat
 				}
-				prob0 := V[t-1][y0] + transP + em_p
+				prob0 := V[t-1][y0] + transP + emP
 				ps0 = append(ps0, &probState{prob: prob0, state: y0})
 			}
 			sort.Sort(sort.Reverse(ps0))