Lucene Analyzers for Buddhist Chinese

This repository contains bricks to process Chinese in Lucene, mainly:

• stopwords
• SC/TC conversion
• Hanzi variants
• Hanzi to Pinyin conversion
• Pinyin syllable tokenizer

Installation

You can install this analyzer from Maven:

    <dependency>
      <groupId>io.bdrc.lucene</groupId>
      <artifactId>lucene-zh</artifactId>
      <version>0.2.0</version>
    </dependency>

Building from source

A compiled Trie is needed in order to build a complete jar, the base command line to build a jar is thus:

mvn clean compile exec:java package

The following options alter the packaging:

• -DincludeDeps=true includes io.bdrc.lucene:stemmer in the produced jar file
• -DperformRelease=true signs the jar file with gpg

Indexing Pipeline

TC ⟾ normalized TC ⟾ SC ⟾ normalized SC ⟾ normalized PY_strict ⟾ normalized PY_lazy

normalized TC/SC: any combination of the following treatments: synonyms, alternatives and stopwords.

normalized PY: PY is lower-cased and split in syllables (to match the general policy of indexing individual ideograms). We call PY_strict Pinyin with tone indication (diacritics or numbers) and PY_lazy Pinyin with no tone indication.

Constructors

ChineseAnalyzer(String indexEncoding,String inputEncoding,boolean stopwords, int variants) 
    indexEncoding- "TC", "SC", "PY_strict" or "PY_lazy"
    inputEncoding- "TC", "SC", "PY_strict" or "PY_lazy"
    stopWords    - true to filter stopwords, false otherwise
    variants     - 0: no variant; 1: synonyms; 2: alternatives; 3: both

ChineseAnalyzer(String profile)

Profiles	inputEncoding	indexEncoding	stopWords	variants
exactTC	TC	TC	false	0
TC	TC	TC	true	3
TC2SC	TC	SC	true	3
TC2PYstrict	TC	PYstrict	true	3
TC2PYlazy	TC	PYlazy	true	3
SC	SC	SC	true	3
SC2PYstrict	SC	PYstrict	true	3
SC2PYlazy	SC	PYlazy	true	3
PYstrict	PYstrict	PYstrict	false	0
PYstrict2PYlazy	PYstrict	PYlazy	false	0
PYlazy	PYlazy	PYlazy	false	0

Components

Tokenizers

StandardTokenizer

Produces ideogram-based tokens(it incorporated the historical Chinese Tokenizer).

WhitespaceTokenizer

Used together with PinyinSyllabifyingFilter in order to avoid giving it big strings.

Filters

PinyinNormalizingFilter (MappingCharFilter)

TODO: when we have more pinyin data or when we know how users type their queries, assess if the normalization is sufficient or not.

TC2SCFilter (TokenFilter)

Leverages Unihan data to replace token content with the SC equivalent.

ZhToPinyinFilter (TokenFilter)

Replaces the token content(TC and SC) with the pinyin transcription.

LazyPinyinFilter (TokenFilter)

Removes tone marks in Pinyin.

LowerCaseFilter (MappingCharFilter)

Used as a pre-processing step for PY indexing

PinyinSyllabifyingFilter (TokenFilter)

Produces syllable-based tokens using PinyinAlphabetTokenizer. Supports both strict and lazy pinyin.

ZhSynonymsFilter (MappingCharFilter)

Leverages Unihan's kSemanticVariant field to index the same variant for all synonyms.

ZhAlternatesFilter (MappingCharFilter)

Leverages Unihan's kZVariant field to index the same variant for stylistic variants of the same ideogram.

Sizes

In the Unihan database for Unicode 10, 88884 codepoints have an entry (adding full ideograms and parts of surrogate pairs). 82829 entries have no information about being TC nor SC, 3037 are specifically TC, 3007 are specifically SC and 11 have information about both TC and SC.

There are 1655 possible syllables in PY and 469 in PY with no diacritics.

Resources

src/main/resources is the output of lucene-zh-data, generated by make, except for pinyin-alphabet.dict, coming from here.

src/main/resources/zh-stopwords.txt is this stop-list

src/main/resources/zh-stopwords_analyzed.txt is the same list as above with the corresponding SC, PYstrict and PYlazy corresponding strings. It was generated using PrettyPrintResult.java.

Licence

The code is Copyright 2018 Buddhist Digital Resource Center, and is provided under Apache License 2.0.

Files in src/main/resources remain under Unicode License, except zh-stopwords.txt, under MIT Licence and pinyin-alphabet.dict, under Apache 2 Licence.