進行情緒分析時，需要有斷詞工具，透過斷字斷詞擷取出關鍵字
再透過關鍵字去判斷為正面或負面情緒，因此還需定義正負評價詞庫
最後透過計算正負關鍵字出現次數來決定該文的正負評
1.中文斷字斷詞
本文採用mmseg4j作為中文的斷字斷詞，會使用mmseg4j原因是可自訂詞庫
可參考
GitHub:http://ift.tt/2nRPbDU
mmseg4j 中文斷詞java 實作:http://ift.tt/2oFGaL9
中文分詞器性能比較:http://ift.tt/SrFUiJ
1.1 詞庫目錄
首先在src\main\resources建立詞庫檔案
詞庫定義如下
1.2 載入多個詞庫
如下方Line 68指定詞庫所在路徑，將會自動載入words***.dic的詞庫檔案(words開頭.dic結尾)
目的為將詞庫分類，較容易管理與擴充，Line 62路徑即為 1.1詞庫目錄
1.3 測試斷詞功能
測試文句:
疾管署今天表示，9月在高雄國際航空站捕到漢他病毒鼠，但沒有即時告知。雖未釀成疫情，疾管署承認疏失，將檢討內部流程並相關處分。
org.iwlp.controller.SemanticAnalysisController
|
@RequestMapping(method=RequestMethod.POST,value="/segment")
@ApiOperation(value = "中文斷詞",notes="輸入中文斷詞處理的文章內容，檢視斷詞結果")
@ApiImplicitParams({
@ApiImplicitParam(name = "text", value = "text", required = true, defaultValue = "這行文字是要被中文斷詞處理的文章，可以從執行結果看斷詞是否成功",dataType = "string", paramType = "query"),
@ApiImplicitParam(name = "fieldType", value = "fieldType", required = false, defaultValue = "ComplexSeg",dataType = "string", paramType = "query"),
@ApiImplicitParam(name = "isRefresh", value = "是否刷新字典檔", required = true, defaultValue = "false",dataType = "boolean", paramType = "query")
})
public Object segment(
@RequestParam(required=true, value = "text") String text,
@RequestParam(required = false, value = "fieldType") FieldTypeEnum fieldType,
@RequestParam(required=true, value = "isRefresh") Boolean isRefresh
) {
String result = null;
Segment segment = new Segment(isRefresh);
if(fieldType == null){
fieldType = FieldTypeEnum.ComplexSeg;
}
try {
result = segment.segWords(text, " | ", fieldType);
} catch (IOException e) {
e.printStackTrace();
}
TextBody textBody = new TextBody();
textBody.setText(text);
return result;
}
|
疾 | 管 | 署 | 今天 | 表示 | 9 | 月 | 在 | 高雄 | 國際航空 | 站 | 捕 | 到 | 漢 | 他 | 病毒 | 鼠 | 但沒 | 有 | 即時 | 告知 | 雖未 | 釀成 | 疫情 | 疾 | 管 | 署 | 承認 | 疏失 | 將 | 檢討 | 內部 | 流程 | 並 | 相關 | 處分
|
1.4 調整詞庫
src\main\resources\data\words-lexicon.dic
最後一行加入自訂詞庫"疾管署", "漢他病毒"
輸出結果
疾管署 | 今天 | 表示 | 9 | 月 | 在 | 高雄 | 國際航空 | 站 | 捕 | 到 | 漢他病毒 | 鼠 | 但沒 | 有 | 即時 | 告知 | 雖未 | 釀成 | 疫情 | 疾管署 | 承認 | 疏失 | 將 | 檢討 | 內部 | 流程 | 並 | 相關 | 處分
1.5 斷詞(分詞)完整程式碼
package org.iwlp.core;
import java.io.IOException;
import java.io.Reader;
import java.io.StringReader;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.HashSet;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.Set;
import org.apache.commons.io.IOUtils;
import org.iwlp.model.FieldTypeEnum;
import org.iwlp.model.result.KeyWordFrequency;
import org.iwlp.model.result.KeyWordsStatistic;
import org.iwlp.utils.MapSortingUtils;
import org.iwlp.utils.StreamUtils;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.chenlb.mmseg4j.ComplexSeg;
import com.chenlb.mmseg4j.Dictionary;
import com.chenlb.mmseg4j.MMSeg;
import com.chenlb.mmseg4j.MaxWordSeg;
import com.chenlb.mmseg4j.Seg;
import com.chenlb.mmseg4j.SimpleSeg;
import com.chenlb.mmseg4j.Word;
import com.google.gson.Gson;
import com.google.gson.GsonBuilder;
/**
* 程式資訊摘要：分詞<P>
* 類別名稱 ：Segment.java<P>
* 程式內容說明：<P>
* 程式修改記錄：<P>
* date：2016年9月8日<P>
*@author A50319
*@version 1.0
*@since 1.0
*/
public class Segment {
private Dictionary dic;
private static final Logger log = LoggerFactory.getLogger(Segment.class);
/**
* Segment建構子
* @param isRefresh : 是否重新刷新Dictionary
*/
public Segment(boolean isRefresh) {
initDictionary(isRefresh);
}
private void initDictionary(boolean isRefresh){
String filePath = StreamUtils.getResourceURL(Segment.class, "data").getFile();
if(isRefresh){
Dictionary.clear(filePath);
log.debug("path:{}", filePath);
log.debug("Dictionary refresh sataus:{}", isRefresh);
}
System.setProperty("mmseg.dic.path", filePath); //這裡可以指定自訂詞庫
dic = Dictionary.getInstance();
}
/**
* 預設SEG
* @return ComplexSeg
*/
private Seg getSeg() {
return new ComplexSeg(dic);
}
private Seg getSeg(FieldTypeEnum fieldType) {
Seg seg = null;
// log.debug("fieldType:{}", fieldType.name());
if(fieldType == FieldTypeEnum.SimpleSeg){
// log.debug("SimpleSeg:{}", FieldTypeEnum.SimpleSeg.name());
seg = new SimpleSeg(dic);
} else if(fieldType == FieldTypeEnum.ComplexSeg){
// log.debug("ComplexSeg:{}", FieldTypeEnum.ComplexSeg.name());
seg = new ComplexSeg(dic);
} else if(fieldType == FieldTypeEnum.MaxWordSeg){
// log.debug("MaxWordSeg:{}", FieldTypeEnum.MaxWordSeg.name());
seg = new MaxWordSeg(dic);
} else{
seg = getSeg();
}
return seg;
}
/**
* 斷詞(預設ComplexSeg分詞器)
* @param text : 內文
* @param wordSpilt : 分割標記文字
* @return String
* @throws IOException
*/
public String segWords(String text, String wordSpilt) throws IOException {
Reader inputReader = new StringReader(text);
StringBuilder sb = new StringBuilder();
Seg seg = getSeg();
MMSeg mmSeg = new MMSeg(inputReader, seg);
Word word = null;
boolean first = true;
while((word=mmSeg.next())!=null) {
if(!first) {
sb.append(wordSpilt);
}
String w = word.getString();
sb.append(w);
first = false;
}
return sb.toString();
}
/**
* 斷詞
* @param text : 內文
* @param wordSpilt : 分割標記文字
* @param fieldType : 分詞器類型
* @return String
* @throws IOException
*/
public String segWords(String text, String wordSpilt, FieldTypeEnum fieldType) throws IOException {
Reader inputReader = new StringReader(text);
StringBuilder sb = new StringBuilder();
Seg seg = getSeg(fieldType);
MMSeg mmSeg = new MMSeg(inputReader, seg);
Word word = null;
boolean first = true;
while((word=mmSeg.next())!=null) {
if(!first) {
sb.append(wordSpilt);
}
String w = word.getString();
sb.append(w);
first = false;
}
return sb.toString();
}
/**
* 斷詞(預設ComplexSeg分詞器)
* @param text : 內文
* @return List<String> words
*/
public List<String> segWords(String text) {
List<String> words = new ArrayList<String>();
Reader inputReader = new StringReader(text);
Seg seg = getSeg();
MMSeg mmSeg = new MMSeg(inputReader, seg);
Word word = null;
try {
while((word=mmSeg.next())!=null) {
words.add(word.getString());
}
} catch (IOException e) {
e.printStackTrace();
} finally{
IOUtils.closeQuietly(inputReader);
}
return words;
}
/**
* 斷詞
* @param text : 內文
* @param fieldType : 分詞器類型
* @return List<String> words
*/
public List<String> segWords(String text, FieldTypeEnum fieldType) {
List<String> words = new ArrayList<String>();
Reader inputReader = new StringReader(text);
Seg seg = getSeg(fieldType);
MMSeg mmSeg = new MMSeg(inputReader, seg);
Word word = null;
try {
while((word=mmSeg.next())!=null) {
words.add(word.getString());
}
} catch (IOException e) {
e.printStackTrace();
} finally{
IOUtils.closeQuietly(inputReader);
}
return words;
}
/**
* 計算關鍵字出現頻率
* @param text : 內文
* @param fieldType : 分詞器類型
* @return Map<String, Integer>
*/
public Map<String, Integer> calculateKeywordsMap(String text, FieldTypeEnum fieldType){
Map<String, Integer> keywords = new LinkedHashMap<String, Integer>();
Map<String, Object> tempKeywords = new HashMap<String, Object>();
List<String> words = segWords(text, fieldType);
for(String keyword : words){
if(tempKeywords.containsKey(keyword)){
int count = ((int) tempKeywords.get(keyword)) + 1;
tempKeywords.put(keyword, count);
} else{
tempKeywords.put(keyword, 1);
}
}
Gson gson = new GsonBuilder().create();
// log.debug("sort before:\n{}",gson.toJson(MapSortingUtils.sortMapByValue(tempKeywords)));
tempKeywords = MapSortingUtils.reverseMap(MapSortingUtils.sortMapByValue(tempKeywords));
// log.debug("sort affter:\n{}",gson.toJson(tempKeywords));
for(Entry<String, Object> entry : tempKeywords.entrySet()){
String key = entry.getKey();
Integer value = (Integer) entry.getValue();
keywords.put(key, value);
// log.debug("key:{}, value:{}",key, value);
}
tempKeywords.clear();
tempKeywords = null;
return keywords;
}
/**
* 計算關鍵字出現頻率
* @param text : 內文
* @param fieldType : 分詞器類型
* @return KeyWordsStatistic
*/
public KeyWordsStatistic calculateKeywords(String text, FieldTypeEnum fieldType){
KeyWordsStatistic ks = new KeyWordsStatistic();
Map<String, Integer> keywords = calculateKeywordsMap(text, fieldType);
ks.setTotal(keywords.size());
List<KeyWordFrequency> kfs = new ArrayList<KeyWordFrequency>();
for(Entry<String, Integer> entry : keywords.entrySet()){
String key = entry.getKey();
Integer value = (Integer) entry.getValue();
KeyWordFrequency kf = new KeyWordFrequency();
kf.setKeyWord(key);
kf.setCount(value);
kfs.add(kf);
}
ks.setResult(kfs);
return ks;
}
/**
* 計算關鍵字出現頻率
* @param text : 內文
* @param fieldType : 分詞器類型
* @param minKeywordLength : 關鍵字最小長度門檻
* @param minFrequency : 關鍵字出現最小次數門檻
* @return KeyWordsStatistic
*/
public KeyWordsStatistic calculateKeywords(String text, FieldTypeEnum fieldType, int minKeywordLength, int minFrequency){
KeyWordsStatistic ks = new KeyWordsStatistic();
Map<String, Integer> keywords = calculateKeywordsMap(text, fieldType);
List<KeyWordFrequency> kfs = new ArrayList<KeyWordFrequency>();
for(Entry<String, Integer> entry : keywords.entrySet()){
String key = entry.getKey();
Integer value = (Integer) entry.getValue();
if(key.length() >= minKeywordLength){
if(value >= minFrequency){
KeyWordFrequency kf = new KeyWordFrequency();
kf.setKeyWord(key);
kf.setCount(value);
kfs.add(kf);
}
}
}
ks.setResult(kfs);
ks.setTotal(ks.getResult().size());
return ks;
}
}
2.關鍵字統計
此功能目的為統計詞庫檔案內關鍵字於文章中出現的頻率，只要斷字斷詞在統計出現次數即可
minKeywordLength:關鍵字最小文字長度，若值為3，則"疾管署 | 承認 | 疏失"，僅出現長度為3的關鍵字"疾管署"
minFrequency:關鍵字出現頻率，篩選關鍵字出現次數
2.1 測試關鍵字統計
org.iwlp.controller.SemanticAnalysisController
|
@RequestMapping(method=RequestMethod.POST,value="/keywords")
@ApiOperation(value = "關鍵字統計",notes="根據文章內容統計關鍵字出現次數")
@ApiImplicitParams({
@ApiImplicitParam(name = "isRefresh", value = "是否刷新字典檔", required = true, defaultValue = "false",dataType = "boolean", paramType = "query"),
@ApiImplicitParam(name = "fieldType", value = "fieldType", required = false, defaultValue = "ComplexSeg",dataType = "string", paramType = "query"),
@ApiImplicitParam(name = "minKeywordLength", value = "關鍵字最小文字長度", required = false, defaultValue = "1",dataType = "integer", paramType = "query"),
@ApiImplicitParam(name = "minFrequency", value = "關鍵字出現頻率", required = false, defaultValue = "1",dataType = "integer", paramType = "query")
})
public KeyWordsStatistic calculateKeywords(
@RequestParam(required=true, value = "isRefresh") Boolean isRefresh,
@RequestParam(required = false, value = "fieldType") FieldTypeEnum fieldType,
@RequestParam(required = false, value = "minKeywordLength") Integer minKeywordLength,
@RequestParam(required = false, value = "minFrequency") Integer minFrequency,
@RequestBody(required=true) final TextBody body
) {
if(minKeywordLength == null){
minKeywordLength = 1;
}
if(minFrequency == null){
minFrequency = 1;
}
AssertUtils.isMoreThan(minKeywordLength, 1);
AssertUtils.isMoreThan(minFrequency, 1);
TextBody textBody = gson.fromJson(gson.toJson(body), TextBody.class);
Segment segment = new Segment(isRefresh);
if(fieldType == null){
fieldType = FieldTypeEnum.ComplexSeg;
}
KeyWordsStatistic kws = segment.calculateKeywords(textBody.getText(), fieldType, minKeywordLength, minFrequency);
return kws;
}
|
{"total":29,"result":[{"keyword":"疾管署","count":2},{"keyword":"高雄","count":1},{"keyword":"9","count":1},{"keyword":"疫情","count":1},{"keyword":"到","count":1},{"keyword":"內部","count":1},{"keyword":"相關","count":1},{"keyword":"在","count":1},{"keyword":"流程","count":1},{"keyword":"並","count":1},{"keyword":"即時","count":1},{"keyword":"承認","count":1},{"keyword":"檢討","count":1},{"keyword":"鼠","count":1},{"keyword":"漢他病毒","count":1},{"keyword":"站","count":1},{"keyword":"國際航空","count":1},{"keyword":"今天","count":1},{"keyword":"捕","count":1},{"keyword":"告知","count":1},{"keyword":"疏失","count":1},{"keyword":"有","count":1},{"keyword":"月","count":1},{"keyword":"將","count":1},{"keyword":"雖未","count":1},{"keyword":"但沒","count":1},{"keyword":"釀成","count":1},{"keyword":"表示","count":1},{"keyword":"處分","count":1}]}
|
3.情緒詞庫
本文中情緒分析採用作者der3318所開發之情緒分析，並加以修改
根據第2章之中文斷詞機制將文章的關鍵字分離出來，並判斷這些關鍵字代表之情緒為正面或負面
經統計後可得情緒分析分數
可參考
SentimentAnalyzer:http://ift.tt/2oG3Yi3
內建採用結巴(jieba)中文斷詞，此處修改為mmseg4j，原因是原先系統已採用mmseg4j
3.1 情緒詞庫目錄
最主要為定義正面詞庫(positive.txt)以及負面詞庫(negative.txt)
positive.txt
negative.txt
3.2 中文斷詞設定
此處故意沒刪除結巴(jieba)中文斷詞之處，請參閱綠色被註解的地方
org.iwlp.core.analyzer.SegChinese
|
package org.iwlp.core.analyzer;
import java.io.IOException;
import java.util.ArrayList;
import org.iwlp.core.Segment;
import org.iwlp.model.FieldTypeEnum;
//import com.huaban.analysis.jieba.JiebaSegmenter;
//import com.huaban.analysis.jieba.JiebaSegmenter.SegMode;
//import com.huaban.analysis.jieba.SegToken;
public class SegChinese {
// a static Segmenter shared by all analyzers
private static SegChinese seg;
//mmseg4j
private Segment segmenter;
public static boolean isRefresh;
//結巴分詞
// protected JiebaSegmenter segmenter;
// public SegChinese() {
// segmenter = new JiebaSegmenter();
// }
public SegChinese(boolean isRefresh) {
segmenter = new Segment(isRefresh);
}
// return the prepared Segmenter. if not found, create one
public static SegChinese getInstance() {
if (seg == null) {
synchronized (SegChinese.class) {
if (seg == null) {
seg = new SegChinese(isRefresh);
return seg;
}
}
}
return seg;
}
public static void removeInstance() {
seg = null;
}
public ArrayList<String> getSegList(String text) throws IOException {
return (ArrayList<String>) segmenter.segWords(text, FieldTypeEnum.ComplexSeg);
}
public String segWords(String text, String wordSpilt) throws IOException {
return segmenter.segWords(text, wordSpilt);
}
// public ArrayList<String> getSegList(String txt) throws IOException {
// ArrayList<String> output = new ArrayList<String>();
// for( SegToken token : segmenter.process(txt, SegMode.INDEX) ) if( !token.word.isEmpty() ) output.add(token.word);
// return output;
// }
//
// public String segWords(String txt, String wordSpilt) throws IOException {
// String output = new String("");
// for( SegToken token : segmenter.process(txt, SegMode.INDEX) ) output += (token.word + wordSpilt);
// return output;
// }
}
|
3.3 情緒分析流程
情緒計算方式為將文章
疾管署今天表示，9月在高雄國際航空站捕到漢他病毒鼠，但沒有即時告知。雖未釀成疫情，疾管署承認疏失，將檢討內部流程並相關處分。衛生福利部疾病管制署今天發布新聞稿表示，9月在高雄國際航空站捕獲錢鼠送實驗室檢驗檢出漢他病毒陽性，當時並未立即通知航空站，日前與航站定期召開衛生小組會議時，才告知並要求加強環境清潔與防鼠措施。
用以設定之中文斷詞進行斷句斷詞
疾管署 | 今天 | 表示 | 9 | 月 | 在 | 高雄 | 國際航空 | 站 | 捕 | 到 | 漢他病毒 | 鼠 | 但沒 | 有 | 即時 | 告知 | 雖未 | 釀成 | 疫情 | 疾管署 | 承認 | 疏失 | 將 | 檢討 | 內部 | 流程 | 並 | 相關 | 處分 | 衛生 | 福利 | 部 | 疾病 | 管制 | 署 | 今天 | 發 | 布 | 新聞稿 | 表示 | 9 | 月 | 在 | 高雄 | 國際航空 | 站 | 捕獲 | 錢鼠 | 送 | 實驗室 | 檢驗 | 檢出 | 漢他病毒 | 陽性 | 當時 | 並未 | 立即 | 通知 | 航空站 | 日前 | 與 | 航 | 站 | 定期 | 召開 | 衛生 | 小組 | 會議 | 時 | 才 | 告知 | 並要 | 求 | 加強 | 環境 | 清潔 | 與 | 防鼠 | 措施
之後再進行正負情緒關鍵字之初始化，即正面詞庫分數定為1分，負面詞庫定為-1分 ，此處將英文字母設定為轉為小寫
package org.iwlp.core.analyzer;
import java.io.BufferedReader;
import java.io.FileReader;
import java.util.ArrayList;
import java.util.HashMap;
import org.apache.commons.io.IOUtils;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
public class SentimentalDictionary {
private static final Logger log = LoggerFactory.getLogger(SentimentalDictionary.class);
// a static dictionary shared by all analyzers
private static SentimentalDictionary dictionary;
// filenames
private static String filenameP = new String("./docs/positive.txt");
private static String filenameN = new String("./docs/negative.txt");
private static String filenameADV = new String("./docs/adv.txt");
// a HashMap holding sentimental words as keys
private HashMap<String, Integer> mydict = new HashMap<String, Integer>();
// a HashMap holding adverbs as keys
private HashMap<String, Boolean> myadv = new HashMap<String, Boolean>();
// return the prepared dictionary. if not found, create one
public static SentimentalDictionary getInstance() {
if (dictionary == null) {
synchronized (SentimentalDictionary.class) {
if (dictionary == null) {
dictionary = new SentimentalDictionary();
dictionary.makeDict();
return dictionary;
}
}
}
return dictionary;
}
// remove the current dictionary due to some setting changes
public static void removeInstance() {
dictionary = null;
}
public static void setFilename(String _filenameP, String _filenameN, String _filenameADV) {
filenameP = _filenameP;
filenameN = _filenameN;
filenameADV = _filenameADV;
}
// add a positive word into dictionary
public synchronized void addPositiveWords(String _string) {
if( mydict.containsKey(_string) ) mydict.put(_string, mydict.get(_string) + 1);
else mydict.put(_string, 1);
}
// add a negative word into dictionary
public synchronized void addNegativeWords(String _string) {
if( mydict.containsKey(_string) ) mydict.put(_string, mydict.get(_string) - 1);
else mydict.put(_string, -1);
}
// get the score of the sentimental word, and return 0 when not found
public int checkWord(String _string) {
if( _string.isEmpty() || !mydict.containsKey(_string) ) return 0;
if( mydict.get(_string) > 0 ) return 1;
return -1;
}
// check if the input word is adv or not
public boolean checkAdv(String _string) {
if( myadv.containsKey(_string) && !_string.isEmpty() ) return true;
return false;
}
// return an ArrayList containing positive words
public ArrayList<String> getPositiveWords() {
ArrayList<String> output_list = new ArrayList<String>();
for( String key : mydict.keySet() ) if( mydict.get(key) > 0 ) output_list.add(key);
return output_list;
}
// return an ArrayList containing negative words
public ArrayList<String> getNegativeWords() {
ArrayList<String> output_list = new ArrayList<String>();
for( String key : mydict.keySet() ) if( mydict.get(key) < 0 ) output_list.add(key);
return output_list;
}
// get the size(numbers of words) of the dictionary
public int getSize() {
return mydict.size() + myadv.size();
}
// print dictionary
public void printDict() {
for( String key : mydict.keySet() ) System.out.println(key + ", " + mydict.get(key));
}
// put the words into the the HashMaps from 3 input files(positive sentimental words, negative sentimental words, adverbs)
public void makeDict() {
try {
// access positive words
String[] filenames = {filenameP, /*"./docs/pos_by_training.txt"*/};
for(String filename : filenames) {
log.debug("Accessing :{}" , filename);
FileReader fr = new FileReader(filename);
BufferedReader br = new BufferedReader(fr);
String tmp = br.readLine();
while(tmp != null) {
//新增正面詞庫(英文自動轉小寫)
addPositiveWords( tmp.trim().toLowerCase() );
tmp = br.readLine();
}
br.close();
}
}
catch (Exception e) {
System.out.println("File of Positive Words Not Found");
e.printStackTrace();
}
try {
// access negative words
String[] filenames = {filenameN/*, "./docs/neg_by_training.txt"*/};
for(String filename : filenames) {
// System.out.println("Accessing " + filename);
log.debug("Accessing:{}", filename);
FileReader fr = new FileReader(filename);
BufferedReader br = new BufferedReader(fr);
String tmp = br.readLine();
while(tmp != null) {
//新增負面詞庫(英文自動轉小寫)
addNegativeWords( tmp.trim().toLowerCase() );
tmp = br.readLine();
}
br.close();
}
}
catch (Exception e) {
System.out.println("File of Negative Words Not Found");
e.printStackTrace();
}
try {
log.debug("Accessing :{}", filenameADV);
// access negative words
FileReader fr = new FileReader(filenameADV);
BufferedReader br = new BufferedReader(fr);
String tmp = br.readLine();
while(tmp != null) {
myadv.put(tmp.trim() , true);
tmp = br.readLine();
}
// br.close();
IOUtils.closeQuietly(br);
IOUtils.closeQuietly(fr);
}
catch (Exception e) {
System.out.println("File of Adverbs Not Found");
e.printStackTrace();
}
}
}
再從斷字後的關鍵字中查找正面情緒詞庫和負面情緒詞庫，並計算出現次數
承認(1) | 疏失(-1) | 檢討(-1) | 處分(-1) | 衛生(1) | 衛生(1) | 清潔(1)
統計所有正負評分
正面關鍵字分數6
負面關鍵字分數8
6-8 = -2, 即本文章情緒分數總分
{
"time": "2 mSec",
"lexicon": {
"positive": 6,
"negative": 8
},
"comment": "Negative",
"result": [
{
"keyword": "衛生",
"value": 2
},
{
"keyword": "承認",
"value": 1
},
{
"keyword": "清潔",
"value": 1
},
{
"keyword": "福利",
"value": 1
},
{
"keyword": "通知",
"value": 1
},
{
"keyword": "毒",
"value": -2
},
{
"keyword": "要求",
"value": -1
},
{
"keyword": "處分",
"value": -1
},
{
"keyword": "檢討",
"value": -1
},
{
"keyword": "疾病",
"value": -1
},
{
"keyword": "沒有",
"value": -1
},
{
"keyword": "疏失",
"value": -1
}
]
}
3.4 測試情緒分析
org.iwlp.controller.SemanticAnalysisController
|
@RequestMapping(method=RequestMethod.POST,value="/sentimentAnalysis")
@ApiOperation(value = "情緒分析",notes="根據文章內容正面與負面關鍵字分析文章正負評價")
@ApiImplicitParams({
// @ApiImplicitParam(name = "text", value = "text", required = true, defaultValue = "這酒店骯髒環境差，服務人員態度不佳，不敢恭維",dataType = "string", paramType = "query"),
@ApiImplicitParam(name = "topN", value = "正負面詞庫數量門檻</br>-1:不限數量,default:10", required = false, defaultValue = "10",dataType = "int", paramType = "query"),
@ApiImplicitParam(name = "isRefresh", value = "是否刷新字典檔", required = true, defaultValue = "false",dataType = "boolean", paramType = "query")
})
public Object sentimentAnalysis(
// @RequestParam(value = "text") String text,
@RequestParam(required = false, value = "topN") Integer topN,
@RequestParam(required=true, value = "isRefresh") Boolean isRefresh,
@RequestBody(required=true) final TextBody body
) {
if(topN == null){
topN = 10;
}
AssertUtils.isMoreThan(topN, -1);
TextBody textBody = gson.fromJson(gson.toJson(body), TextBody.class);
Sentiment sentiment = new Sentiment();
String positiveDict = StreamUtils.getResourceURL(Segment.class, "sa/positive.txt").getFile();
String negativeDict = StreamUtils.getResourceURL(Segment.class, "sa/negative.txt").getFile();
String advDict = StreamUtils.getResourceURL(Segment.class, "sa/adv.txt").getFile();
String trainingFile = StreamUtils.getResourceURL(Segment.class, "sa/training.txt").getFile();
String trainingAnswer = StreamUtils.getResourceURL(Segment.class, "sa/answer.txt").getFile();
// String option = StreamUtils.getResourceURL(Segment.class, "sa/opinion.txt").getFile();
// String result = StreamUtils.getResourceURL(Segment.class, "sa/sa-result.txt").getFile();
// log.debug("positiveDict:{}",positiveDict);
// log.debug("negativeDict:{}",negativeDict);
// log.debug("advDict:{}",advDict);
// log.debug("trainingFile:{}",trainingFile);
// log.debug("trainingAnswer:{}",trainingAnswer);
// log.debug("option:{}",option);
// log.debug("result:{}",result);
SentimentAnalyzer.setDictionary(positiveDict, negativeDict, advDict);
SentimentAnalyzer.setTrainingData(trainingFile, trainingAnswer);
//set PMI-SO Rate, deciding how strictly the analyzer chooses words (default = 3.0)
SentimentAnalyzer.setSORate(3.0);
//set the number of threads available for the analyzer (default = 4)
SentimentAnalyzer.setNTHREADS(4);
//create the analyzer wth I/O filename (default= ./docs/opinion.txt, ./result.txt)
// String posByTrainingPath, String negByTrainingPath
// String posByTrainingPath = StreamUtils.getResourceURL(Segment.class, "sa/positive.txt").getFile();
// String negByTrainingPath = StreamUtils.getResourceURL(Segment.class, "sa/negative.txt").getFile();
// SentimentAnalyzer sa = new SentimentAnalyzer(option, result, posByTrainingPath, negByTrainingPath, isRefresh);
SentimentAnalyzer sa = new SentimentAnalyzer(isRefresh);
sentiment = sa.calculateSentiment(textBody.getText(), topN);
// sa.workAndWriteFile();
return sentiment;
}
|
{
"time": "2 mSec",
"lexicon": {
"positive": 6,
"negative": 8
},
"comment": "Negative",
"result": [
{
"keyword": "衛生",
"value": 2
},
{
"keyword": "承認",
"value": 1
},
{
"keyword": "清潔",
"value": 1
},
{
"keyword": "福利",
"value": 1
},
{
"keyword": "通知",
"value": 1
},
{
"keyword": "毒",
"value": -2
},
{
"keyword": "要求",
"value": -1
},
{
"keyword": "處分",
"value": -1
},
{
"keyword": "檢討",
"value": -1
},
{
"keyword": "疾病",
"value": -1
},
{
"keyword": "沒有",
"value": -1
},
{
"keyword": "疏失",
"value": -1
}
]
}
|
4.API DOC
http://ift.tt/2oG3qc8
|
5.專案配置檔
Gradle configuation
build.gradle
|
buildscript {
ext {
springBootVersion = '1.4.2.RELEASE'
}
repositories {
mavenCentral()
}
dependencies {
classpath("org.springframework.boot:spring-boot-gradle-plugin:${springBootVersion}")
}
}
apply plugin: 'java'
apply plugin: 'eclipse-wtp'
apply plugin: 'org.springframework.boot'
apply plugin: 'war'
war {
baseName = 'iwlp-sa'
version = '0.0.1-SNAPSHOT'
}
sourceCompatibility = 1.8
targetCompatibility = 1.8
repositories {
mavenCentral()
}
configurations {
providedRuntime
}
dependencies {
compile('org.springframework.boot:spring-boot-starter-web')
runtime('mysql:mysql-connector-java')
providedRuntime('org.springframework.boot:spring-boot-starter-tomcat')
testCompile('org.springframework.boot:spring-boot-starter-test')
//swagger
compile group: 'io.springfox', name: 'springfox-swagger2', version: '2.5.0'
compile group: 'io.springfox', name: 'springfox-swagger-ui', version: '2.5.0'
compile group: 'com.restfb', name: 'restfb', version: '1.31.0'
// http://ift.tt/2o4C7vm
compile group: 'org.jsoup', name: 'jsoup', version: '1.9.2'
//FOR CrawlerPack
compile group: 'com.googlecode.juniversalchardet', name: 'juniversalchardet', version: '1.0.3'
//中文斷詞
// http://ift.tt/2nRHhdx
compile group: 'com.chenlb.mmseg4j', name: 'mmseg4j-core', version: '1.10.0'
compile group: 'com.chenlb.mmseg4j', name: 'mmseg4j-analysis', version: '1.9.1'
compile group: 'com.chenlb.mmseg4j', name: 'mmseg4j-core-with-dic', version: '1.8.6'
compile group: 'com.google.code.gson', name: 'gson', version: '2.7'
compile group: 'commons-io', name: 'commons-io', version: '2.5'
compile group: 'commons-lang', name: 'commons-lang', version: '2.6'
compile group: 'org.apache.commons', name: 'commons-vfs2', version: '2.1'
//http client
compile group: 'org.apache.httpcomponents', name: 'httpcore', version: '4.4.5'
compile group: 'org.apache.commons', name: 'commons-compress', version: '1.12'
compile group: 'commons-httpclient', name: 'commons-httpclient', version: '3.1'
compile group: 'org.apache.httpcomponents', name: 'httpclient', version: '4.5.2'
compile group: 'org.apache.httpcomponents', name: 'httpmime', version: '4.5.2'
}
springBoot {
mainClass = "org.iwlp.IwlpSaApplication"
}
|
Spring Boot configuation
| application.properties |
server.contextPath=/iwlp-sa/
#logging
logging.pattern.console=%d{HH:mm:ss.SSS} [%thread] %-5level %logger{26}.%M [%line] - %msg%n%rEx
logging.level.org.springframework.web=WARN
logging.level.root=WARN
logging.level.org.iwlp.config=DEBUG
logging.level.org.iwlp.controller=DEBUG
logging.level.org.iwlp.core=DEBUG
endpoints.jmx.domain=iwlp-sa
endpoints.jmx.uniqueNames=true
|
6.專案下載
Tags: Spring, Spring Boot, Sentiment Analyzer, IFTTT-SYNC
November 30, 2016 at 02:23PM
Open in Evernote
