Method for automatic indexing and searching word and word attributes in Chinese text

Abstract

The invention relates to a word in the Chinese text and its relevant attribute automatically indexing and searching method. The method encodes the word attribute and automatically translates the user's word stock into a built-in one. Segment the words of the user's language database automatically with the built-in word stock to generate the word segmenting result. Then the word index and the word's attribute index are generated automatically. Automatically search the user's language database to get the searching result with the user's searching condition, aided by the word attribute index, the word index and the word segmenting result. The invention realizes that the automatically indexing and searching a word and its attribute can be done to any unknown language data, based on any word stock and any word attribute architecture. The invention eliminates the language database marking operations, and improves the efficiency of work.

Description

1. Technical field [0001] The invention relates to a text retrieval technology, in particular to a method for automatic indexing and retrieval of words and word attributes in Chinese texts. 2. Background technology [0002] Personnel engaged in Chinese language teaching, language research, dictionary compilation and language engineering (such as automatic machine translation, automatic reading aloud, automatic speech recognition, etc.) need to accumulate a large number of language examples related to words and word attributes, such as several words behind "because" It is an example of "so", an example of an adjective followed by an adverb, and an example of a verb after the preposition "ba", etc. It is used for language example collection, language phenomenon statistics, and language law induction. [0003] In the past, the accumulation, statistics and rule induction of language examples mainly relied on manual copying of cards. With the development and popularization of co...

AI Technical Summary

Problems solved by technology

[0010] The second is that the search objects are too limited

Due to the large manpower consumption, it is impossible to process a large-scale corpus, so the retrievable objects are very limited

For example, if the corpus of the People's Daily in 1998 has been processed, the language phenomena related to words and word attributes in the People's Daily in 1998 can be retrieved, but those in 1999 cannot be retrieved. As for "Dream of Red Mansions", "Camel "Xiangzi" and other novels, Taiwan and Hong Kong corpus, as long as the corpus has not been processed, it cannot be retrieved

Therefore, this method is far from meeting the wide range of needs of users.

[0011] The third is that the word attribute system and word attribute labeling are too rigid

However, once the corpus tagging project starts, the attribute system cannot be changed. After the corpus tagging is completed, which token and what attribute are also determined, so that it cannot meet the different needs of different people.

[0012] The fourth is that the accuracy is not very high

On the other hand, using the existing technology, word attributes are automatically disambiguated and manually corrected, and the remaining errors will also cause inaccurate attribute retrieval results, and what is more serious is that the retrieval results cannot be guaranteed to be complete.

Images