From 6d43674e6b224281e43ccefc87224a7ba2fbb99a Mon Sep 17 00:00:00 2001 From: Paul Baltescu Date: Sun, 10 Mar 2013 01:01:01 +0000 Subject: Added comments. Hooray! --- extractor/vocabulary.h | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) (limited to 'extractor/vocabulary.h') diff --git a/extractor/vocabulary.h b/extractor/vocabulary.h index 03c7dc66..c8fd9411 100644 --- a/extractor/vocabulary.h +++ b/extractor/vocabulary.h @@ -9,16 +9,33 @@ using namespace std; namespace extractor { +/** + * Data structure for mapping words to word ids. + * + * This strucure contains words located in the frequent collocations and words + * encountered during the grammar extraction time. This dictionary is + * considerably smaller than the dictionaries in the data arrays (and so is the + * query time). Note that this is the single data structure that changes state + * and needs to have thread safe read/write operations. + * + * Note: For an experiment using different vocabulary instances for each thread, + * the running time did not improve implying that the critical regions do not + * cause bottlenecks. + */ class Vocabulary { public: virtual ~Vocabulary(); + // Returns the word id for the given word. virtual int GetTerminalIndex(const string& word); + // Returns the id for a nonterminal located at the given position in a phrase. int GetNonterminalIndex(int position); + // Checks if a symbol is a nonterminal. bool IsTerminal(int symbol); + // Returns the word corresponding to the given word id. virtual string GetTerminalValue(int symbol); private: -- cgit v1.2.3