Wayne Xin created LUCENE-6736:
---------------------------------
Summary: SmartChineseAnalyzer chops English tokens in a
chinese-english mixed sentence.
Key: LUCENE-6736
URL: https://issues.apache.org/jira/browse/LUCENE-6736
Project: Lucene - Core
Issue Type: Bug
Components: modules/analysis
Affects Versions: 5.1
Environment: linux Java 1.7
Reporter: Wayne Xin
I am new with Lucene Analyzer. The following code has predefined the sentence
in "testStr":
String testStr = "女单方面，王适娴second seed和头号种子卫冕冠军西班牙选手马林first
seed同处1/4区，3号种子李雪芮和韩国选手Korean player成池铉处在2/4区，不过成池铉先要过日本小将(Japanese
player)奥原希望这关。下半区，6号种子王仪涵若想晋级决赛secure position. congratulations.";
The printed tokenized result is:
女 单 方面 王 适 娴 second seed 和 头号 种子 卫冕 冠军 西班牙 选手 马 林 first seed 同 处 1 4 区 3 号 种子
李 雪 芮 和 韩国 选手 korean player 成 池 铉 处在 2 4 区 不过 成 池 铉 先 要 过 日本 小将 japanes player
奥 原 希望 这 关 下 半 区 6 号 种子 王 仪 涵 若 想 晋级 决赛 secur posit congratul
As you can see some long English tokens such as Japanese, position and
congratulations are cut short in the tokenization process. I hope I didn't use
it wrong.
Test code:
private static void testChineseTokenizer() {
String testStr = "女单方面，王适娴second seed和头号种子卫冕冠军西班牙选手马林first
seed同处1/4区，3号种子李雪芮和韩国选手Korean player成池铉处在2/4区，不过成池铉先要过日本小将(Japanese
player)奥原希望这关。下半区，6号种子王仪涵若想晋级决赛secure position. congratulations.";
Analyzer analyzer = new SmartChineseAnalyzer();
List<String> result = new ArrayList<String>();
StringReader sr = new StringReader(testStr);
try {
TokenStream stream = analyzer.tokenStream(null,sr);
CharTermAttribute cattr =
stream.addAttribute(CharTermAttribute.class);
stream.reset();
while (stream.incrementToken()) {
String token = cattr.toString();
result.add(token);
}
stream.end();
stream.close();
sr.close();
analyzer.close();
stream = null;
for (String tok: result) {
System.out.print(" " + tok);
}
System.out.println();
}
catch(IOException e) {
// not thrown b/c we're using a string reader...
}
}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
