如何使用Lucene在句子中搜索关键字

Question

I receive vendor name and their address as string which I index in the Lucene like this, , , , . Now, from another servlet I receive text which has vendor names and the address. Example, "I have problem in using the credit card, xxxxx, in the shop , ", or, "my credit card is declined in the shop , ". I remove stop words like I, the, in, problem, shop, etc and have a clean text with " ". I need to find all those vendor(s) from the vendor_name or area inside the given text.

This is how I index the vendor details; every line in the file is a vendor and their details separated by comma. , , ,

FieldType keywordFieldType = new FieldType();
        keywordFieldType.setStored(true);
        keywordFieldType.setIndexed(true);
        keywordFieldType.setTokenized(false);
        writer = new IndexWriter(dir, iwc);
        BufferedReader reader = new BufferedReader(new FileReader(
                VENDOR_DETAILS));
        String line = reader.readLine();
        while (line != null) {
            Document document = new Document();
            document.add(new Field("content", line.toLowerCase(),
                    keywordFieldType));
            writer.addDocument(document);
            line = reader.readLine();
        }
        writer.commit();

This is how I search the index,

QueryParser queryParser = new QueryParser(VERSION, "content",
            new WhitespaceAnalyzer(VERSION));

    String special = "content:" + stringToQuery.trim();
    try {
        if (searcherManager == null) {
            searcherManager = new SearcherManager(
                    FSDirectory.open(new File(INDEX_DIRECTORY)),
                    new SearcherFactory());
        }
        searcher = searcherManager.acquire();
        TopDocs docs = searcher.search(queryParser.parse(special), 100);
        int hitCount = docs.totalHits;

How do I query the Lucene to search the above requirement? What type of Query should I use to find the vendor details inside the given text?

Answer 1

You are adding you documents without tokenization, but are tokenizing the query, so you have a mismatch in your analysis at query vs index time. Since the field appears to be free text, tokenizing it is important to effective searching. Rather than specifying a FieldType at all, I would recommend just using TextField . You can use WhitespaceTokenizer on both, as you do at query time, but I would consider StandardAnalyzer as a better starting point.