find

Documentation

Overview

Functions for finding documents by full text search

Functions for parsing a search query

Index

• Variables
• func Bigrams(terms []string) []string
• func LoadColMap(r io.Reader) (*map[string]string, error)
• func LoadDocInfo(r io.Reader) (*map[string]DocInfo, *map[string]DocInfo)
• type BM25Score
• type Collection
• type DictQueryParser
  • func (parser DictQueryParser) ParseQuery(query string) []TextSegment
• type DocFinder
  • func NewDocFinder(tfDocFinder TermFreqDocFinder, titleFinder TitleFinder) DocFinder
• type DocInfo
• type Document
  • func (doc Document) String() string
• type QueryParser
  • func NewQueryParser(dict map[string]*dicttypes.Word) QueryParser
• type QueryResults
• type TermFreqDocFinder
  • func NewMysqlDocFinder(ctx context.Context, database *sql.DB) TermFreqDocFinder
• type TextSegment
• type TitleFinder
  • func NewFileTitleFinder(colMap *map[string]string, dInfoCN, docMap *map[string]DocInfo) TitleFinder
  • func NewMysqlTitleFinder(ctx context.Context, database *sql.DB, docMap *map[string]DocInfo) (TitleFinder, error)

Variables

var WEIGHT = []float64{0.080, 2.327, 3.040} // [BM25 words, BM25 bigrams, bit vector]

From logistic regression

Functions

func Bigrams

func Bigrams(terms []string) []string

Bigrams constructs a slice of bigrams from pairs of terms

func LoadColMap

func LoadColMap(r io.Reader) (*map[string]string, error)

LoadColMap gets the list of titles of collections in the corpus key: gloss_file, value: title

func LoadDocInfo

func LoadDocInfo(r io.Reader) (*map[string]DocInfo, *map[string]DocInfo)

Load title info for all documents

Types

type BM25Score

type BM25Score struct {
	Document      string
	Collection    string
	Score         float64
	BitVector     float64
	ContainsTerms string
}

type Collection

type Collection struct {
	GlossFile, Title string
}

type DictQueryParser

type DictQueryParser struct {
	Tokenizer *tokenizer.DictTokenizer[*dicttypes.Word]
}

func (DictQueryParser) ParseQuery

func (parser DictQueryParser) ParseQuery(query string) []TextSegment

The method for parsing the query text in this function is based on dictionary lookups

type DocFinder

type DocFinder interface {
	FindDocuments(ctx context.Context, dictSearcher dictionary.ReverseIndex,
		parser QueryParser, query string, advanced bool) (*QueryResults, error)
	FindDocumentsInCol(ctx context.Context, dictSearcher dictionary.ReverseIndex,
		parser QueryParser, query, col_gloss_file string) (*QueryResults, error)
}

DocFinder finds documents.

func NewDocFinder

func NewDocFinder(tfDocFinder TermFreqDocFinder, titleFinder TitleFinder) DocFinder

NewDocFinder creates and initializes an implementation of the DocFinder interface

type DocInfo

type DocInfo struct {
	CorpusFile, GlossFile, Title, TitleCN, TitleEN, CollectionFile, CollectionTitle string
}

type Document

type Document struct {
	GlossFile, Title, CollectionFile, CollectionTitle, ContainsWords string
	ContainsBigrams                                                  string
	SimTitle, SimWords, SimBigram, SimBitVector, Similarity          float64
	ContainsTerms                                                    []string
	MatchDetails                                                     fulltext.MatchingText
	TitleCNMatch                                                     bool
}

func (Document) String

func (doc Document) String() string

For printing out retrieved document metadata

type QueryParser

type QueryParser interface {
	ParseQuery(query string) []TextSegment
}

Parses input queries into a slice of text segments

func NewQueryParser

func NewQueryParser(dict map[string]*dicttypes.Word) QueryParser

Creates a QueryParser

type QueryResults

type QueryResults struct {
	Query, CollectionFile        string
	NumCollections, NumDocuments int
	Collections                  []Collection
	Documents                    []Document
	Terms                        []TextSegment
	SimilarTerms                 []TextSegment
}

type TermFreqDocFinder

type TermFreqDocFinder interface {
	FindDocsTermFreq(ctx context.Context, terms []string) ([]BM25Score, error)
	FindDocsBigramFreq(ctx context.Context, bigrams []string) ([]BM25Score, error)
	FindDocsTermCo(ctx context.Context, terms []string, col string) ([]BM25Score, error)
	FindDocsBigramCo(ctx context.Context, bigrams []string, col string) ([]BM25Score, error)
}

DocFinder finds documents.

func NewMysqlDocFinder

func NewMysqlDocFinder(ctx context.Context, database *sql.DB) TermFreqDocFinder

type TextSegment

type TextSegment struct {
	QueryText string
	DictEntry dicttypes.Word
	Senses    []dicttypes.WordSense
}

A text segment contains the QueryText searched for and possibly a matching dictionary entry. There will only be matching dictionary entries for Chinese words in the dictionary. Non-Chinese text, punctuation, and unknown Chinese words will have nil DictEntry values and matching values will be included in the Senses field.

type TitleFinder

type TitleFinder interface {
	CountCollections(ctx context.Context, query string) (int, error)
	FindCollections(ctx context.Context, query string) []Collection
	FindDocsByTitle(ctx context.Context, query string) ([]Document, error)
	FindDocsByTitleInCol(ctx context.Context, query, col_gloss_file string) ([]Document, error)
	ColMap() *map[string]string
	DocMap() *map[string]DocInfo
}

func NewFileTitleFinder

func NewFileTitleFinder(colMap *map[string]string, dInfoCN, docMap *map[string]DocInfo) TitleFinder

NewDocTitleFinder initializes a DocTitleFinder implementation Params

infoCache: key to the map is the Chinese part of the title

func NewMysqlTitleFinder

func NewMysqlTitleFinder(ctx context.Context, database *sql.DB, docMap *map[string]DocInfo) (TitleFinder, error)