/**
* @author dengxiang.liu
* @description update jieba user dictionary with securities and themes just came up recently;
* should be called before getThemeSecPair methods;
* @see
*/
public void updateLocalJiebaDictWithSecAndTheme() {
String jiebaUserDictPath = JiebaUserDictPath;
logger.info("jieba user dictionary file path: {}", jiebaUserDictPath);
try {
List<DatayesdbpSecurity> securityList = datayesdbpMapper.getSecurityList();
Map<String, String> tokenMap = FileIOUtil.readJiebaDict(jiebaUserDictPath);
logger.info("original jieba dictionary size: {}", tokenMap == null ? 0 : tokenMap.size());
Integer addedNewsWordsCount = 0;
for (DatayesdbpSecurity datayesdbpSecurity : securityList) {
if (!tokenMap.containsKey(datayesdbpSecurity.getSecShortName())) {
tokenMap.put(datayesdbpSecurity.getSecShortName(), "n");
addedNewsWordsCount++;
}
}
List<BigdataTheme> bigdataThemeList = bigdataMapper.getThemeList();
for (BigdataTheme bigdataTheme : bigdataThemeList) {
if (!tokenMap.containsKey(bigdataTheme.getThemeName())) {
tokenMap.put(bigdataTheme.getThemeName(), "n");
}
}
FileIOUtil.writeJiebaDict(jiebaUserDictPath, tokenMap);
logger.info("Add {} new words into jieba dictionary.", addedNewsWordsCount);
} catch (Exception e) {
logger.error("error occurs when update jieba dictionary, {}", e.toString());
}
}
/**
* @author dengxiang.liu
* @return current theme security pairs in database; @Desc get QAed and active theme-security
* relationships from database;
*/
public Map<String, Set<String>> getCurrentThemeSecurityPair() {
List<BigdataThemeSecRel> themeSecPairList = bigdataMapper.getThemeSecurityPairList();
Map<String, Set<String>> themeSecSetMap = new HashMap<String, Set<String>>();
for (BigdataThemeSecRel bigdataThemeSecRel : themeSecPairList) {
Long themeID = bigdataThemeSecRel.getThemeID();
String themeName = bigdataThemeSecRel.getThemeName();
String tickerSymbol = bigdataThemeSecRel.getTickerSymbol();
if (!themeSecSetMap.containsKey(themeID))
themeSecSetMap.put(themeName, new HashSet<String>());
themeSecSetMap.get(themeID).add(tickerSymbol);
}
return themeSecSetMap;
}
/**
* @Author: Dengxiang.Liu @Return: Map<Long, List<ThemeSecurityPair>>; a map from themeID to it's
* ThemeSecurityPair List; @Desc: the result will be QA before insert into Database;
*/
public Map<String, List<ThemeSecurityObjPair>> getThemeSecPairToBeQA() {
logger.info("start getThemeSecPairToBeQA();");
Map<String, Set<String>> curThemeSecPairMap = getCurrentThemeSecurityPair();
List<DatayesdbpSecurity> securityList = datayesdbpMapper.getSecurityList();
/* handle new themes */
List<BigdataTheme> bigdataThemeList = bigdataMapper.getThemeList(); // all themes list;
logger.info(
"all themes list size: {};", bigdataThemeList == null ? 0 : bigdataThemeList.size());
List<BigdataTheme> unhandledThemeList = getUnhandledThemeList(bigdataThemeList);
logger.info(
"unhandled themes list size: {};",
unhandledThemeList == null ? 0 : unhandledThemeList.size());
Set<BigdataTheme> unhandledThemeSet =
(unhandledThemeList == null || unhandledThemeList.isEmpty())
? null
: new HashSet<BigdataTheme>(unhandledThemeList);
Map<String, List<ThemeSecurityObjPair>> newsThemeSecPairMapUnhandled =
textAnalizer.findThemesSecPair(unhandledThemeSet, securityList, false);
Map<String, List<ThemeSecurityObjPair>> crawlThemeSecPairMapUnhandled =
mongoDataAnalizer.findThemeSecPair(unhandledThemeSet, securityList, false);
Map<String, List<ThemeSecurityObjPair>> newsThemeSecPairMapToBeQAUnhandled =
getDiffThemeSecPairBetween(curThemeSecPairMap, newsThemeSecPairMapUnhandled);
Map<String, List<ThemeSecurityObjPair>> crawlThemeSecPairMapToBeQAUnhandled =
getDiffThemeSecPairBetween(curThemeSecPairMap, crawlThemeSecPairMapUnhandled);
/* handle schedule themes: schedule-themes = all-themes - unhandled-themes*/
List<BigdataTheme> scheduleThemeList =
getSchedualThemeList(bigdataThemeList, unhandledThemeList);
Set<BigdataTheme> scheduleThemeSet =
new HashSet<BigdataTheme>(
scheduleThemeList == null ? new ArrayList<BigdataTheme>() : scheduleThemeList);
logger.info(
"schedule themes Set size: {}", scheduleThemeSet == null ? 0 : scheduleThemeSet.size());
Map<String, List<ThemeSecurityObjPair>> newsThemeSecPairMapSchedule =
textAnalizer.findThemesSecPair(scheduleThemeSet, securityList, true);
Map<String, List<ThemeSecurityObjPair>> crawlThemeSecPairMapSchedule =
mongoDataAnalizer.findThemeSecPair(scheduleThemeSet, securityList, true);
Map<String, List<ThemeSecurityObjPair>> newsThemeSecPairMapToBeQASchedule =
getDiffThemeSecPairBetween(curThemeSecPairMap, newsThemeSecPairMapSchedule);
Map<String, List<ThemeSecurityObjPair>> crawlThemeSecPairMapToBeQASchedule =
getDiffThemeSecPairBetween(curThemeSecPairMap, crawlThemeSecPairMapSchedule);
/* integrate theme-security pair maps */
Map<String, List<ThemeSecurityObjPair>> globalThemeSecPairMapToBeQA =
new HashMap<String, List<ThemeSecurityObjPair>>();
IntegrateMaps(globalThemeSecPairMapToBeQA, newsThemeSecPairMapToBeQAUnhandled);
IntegrateMaps(globalThemeSecPairMapToBeQA, newsThemeSecPairMapToBeQASchedule);
IntegrateMaps(globalThemeSecPairMapToBeQA, crawlThemeSecPairMapToBeQAUnhandled);
IntegrateMaps(globalThemeSecPairMapToBeQA, crawlThemeSecPairMapToBeQASchedule);
logger.info("finished getThemeSecPairToBeQA();");
return globalThemeSecPairMapToBeQA;
}