JPS63276630A - Data base system for retrieving designation similarity - Google Patents

Abstract

PURPOSE:To execute an appropriate designation similarity by providing a voice input device, a phoneme converter, and a similar phoneme string retrieving device, retrieving the corresponding information which is brought to a KANA (Japanese syllabary) train notation, from a phoneme string and executing a designation similarity retrieval. CONSTITUTION:In case of executing a designation similarity retrieval, a retriever utters a designation which goes to a retrieval key word, and inputs it to a voice input device 4. A voice which is inputted through this device 4 is changed to a phoneme string notation by a phoneme string converter 5. Also, in a data base body device 3, in addition to conventional information which is brought to an KANA train notation, etc., information which is brought to an phoneme string notation is also accumulated. A phoneme string retrieving device 6 outputted from this device 5 retrieves a phoneme string accumulated in the data base body device 3 through a data base managing device 2 and performs a designation similarity retrieval.

Description

[Detailed Description of the Invention] Industrial Field of Application The present invention is based on the Trademark Examination Standards J
(1981, Japan Institute of Invention and Innovation), which is used to search for registered trademarks with similar names, etc., and which has a function of searching for similar names.

Prior Art Conventional database system example for name similarity search Eva,
Huff: / To magazine (Vo135 No, 9 p4447
The database system introduced in ``Trademark Search System Brandy'' by Matsu 1) is comprised of a kana manual terminal device 1, a database management device 2, and a database main device 3, as shown in FIG. A database searcher inputs search conditions and commands using a character string from the kana input terminal device 1. In accordance with this, the database management device 2 performs a database search using the kana character string. The database main unit 3 stores kana character strings to be searched. When searching for a kana character string that is phonetically similar to a certain kana character string, that is, similar to the pronunciation, from among the kana character strings stored in the database main unit 3, the kana human power terminal calculates the number of characters that differ from the input kana character string. A search is performed by specifying from device 1.

Problems to be Solved by the Invention In a conventional database system having a pronunciation similarity search function using kana character strings, when performing a pronunciation similarity search, it was only possible to perform a search based on a similarity measure between kana character strings. For example, when searching for a kana character string similar to "Ding Tokkyo",
The searcher enters the kana character string "Totsukyo" and instructs the data management device 2 via the kana human power terminal device 1 to search for a character string that differs by two characters. I got search results for the string "Tokyo". However, this search also outputs kana character strings that are difficult to consider as similar to pronunciations, such as "Menkyo." As described above, since a similarity search was performed between kana character strings when originally a similarity search should have been performed between phoneme strings of pronunciation sounds, an appropriate search could not be performed.

Means for solving the problem: In addition to the conventional database system configuration, a voice input device, a phoneme string conversion device that converts input speech into a phoneme string, and a similar phoneme string search device that searches for a phoneme string similar to the input speech. , and performs a pronunciation similarity search by searching for information expressed in a corresponding kana string from a phoneme string.

Function: In addition to the search function of conventional systems, it becomes possible to search for similarities between phoneme strings, and more appropriate pronunciation similarity, that is, similarity search between pronunciation sounds.

Embodiment FIG. 1 shows a database system for name similarity search according to an embodiment of the present invention. In Figure 1, a conventional kana manual terminal device 1, a database management device 2, and a database main device 3 are shown.
In addition, it has a voice input device 4, a phoneme string conversion device 5, and a similar phoneme string search device 6. The phoneme string conversion device 5 and the similar phoneme string search device 6 are realized using a method often used in speech recognition systems, in which input speech is converted into a phoneme string and those similar to the phoneme string are used as recognition candidates. As a reference for speech recognition systems, IEEJ Information Processing Study Group Materials (I
P-85-3r Speaker-independent/Multi-word common speech recognition system J 1985, Futaya 1) Katsuyuki). When performing a pronunciation similarity search, the searcher utters a pronunciation as a search keyword and inputs it into the voice input device 4. Speech input via the voice input device 4 is converted into phoneme string notation by the phoneme string conversion device 5. In addition to conventional information expressed in kana strings, the database main unit 3 also stores information expressed in phoneme strings. The similar phoneme string search device 6 that receives the phoneme string notation output from the phoneme string conversion device 5 searches the phoneme strings stored in the database main body device 3 via the database management device 2, and retrieves the search results. Output.

Effects of the Invention As described above, by using the database system having the configuration of the present invention, it becomes possible to perform a pronunciation similarity search between phoneme strings, and a more appropriate pronunciation similarity search becomes possible.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a database system for name similarity search according to an embodiment of the present invention, and FIG. 2 is a block diagram of a conventional database system. DESCRIPTION OF SYMBOLS 1... Kana manual terminal device, 2... Database management device, 3... Database main unit, 4... Voice input device, 5... Phoneme string conversion device, 6... Similar phoneme string search Device. Name of agent: Patent Attorney Toshio Tsuchinakao and 1 other person Figure 1 Figure 2! ?

Claims (1)

[Claims] A database main body that stores information expressed in kana strings and corresponding phoneme strings, a voice input device for inputting voice, and a phoneme string conversion device that converts the voice input to the voice input device into a phoneme string. , a similar phoneme string search device that searches for kana notation information corresponding to a phoneme string similar to the phoneme string output from the phoneme string conversion device; and a display device that displays the information searched by the search device. Search database system.