JPH11328195A - Character string retrieving device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a character string retrieving device capable of speedily retrieving a document at a high speed in which a designated character string is included. SOLUTION: This device is provided with an input device 32 inputting a character string being a retrieving object, a ward dictionary 32 storing words, a storing part 30 provided with an unnecessary word dictionary 34 storing word information which is not preferable to coincide with a retrieving result and with a connection information storing part 36 storing connection information showing the strength of connection between words, a character string analyzing part 20 dividing the character string inputted from the device 10 into words based on the word information 32 and the unnecessary word information 34 and making the words be a compound word based on the connection information and a retrieving processing part 40 retrieving the compound word from a document data base 50.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character string searching device, and more particularly to a character string searching device for searching a specific character string from an arbitrary character string.

[0002]

2. Description of the Related Art With the advancement of a highly information-oriented society, a technique for quickly searching for a predetermined character string from a database storing a huge amount of data using a computer is very important. For example, in patent information search, the quality and quantity of documents to be searched depend on the search method.

[0003] Therefore, when performing a search, the searcher must specify an appropriate keyword to perform the search. In general, the most basic search method is to narrow down the search targets sequentially using, for example, an AND search or an OR search for a short keyword composed of one word.

[0004]

Conventionally, when a long character string is searched, a processing method (n
In the case of character separation, morphological analysis, etc.), the processing after the separation is performed by performing an OR search and performing its function.
Therefore, as a search target, a document that includes even one word in the document is targeted, and it is necessary to further narrow down a search for documents that are semantically close from an enormous amount of documents. Therefore, conventionally, instead of a long character string search, it is a simple search of a plurality of words, which requires a long search time and has a problem that the search is too time-consuming and troublesome. Also, even if a document that matches more words is retrieved, the connection between words is not taken into account, so a document that has no meaning at all is retrieved, and a valid document that includes the specified keyword is deleted. There is a problem that it is buried in the literature of.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and has as its object to provide a character string search device capable of searching for a document containing a specified character string quickly and at high speed.

[0006]

In order to solve the above-mentioned problems, the present invention preferably provides an input device for inputting a character string to be searched, a word dictionary storing words, and a match as a search result. Unnecessary word dictionary that stores missing word information,
A storage unit having a connection information storage unit that stores connection information indicating the strength of connection between words, and, based on the word information and the unnecessary word information, divides a character string input from the input device into words, A character string analysis unit that uses the word as a compound word based on the connection information and a search processing unit that searches the document database for the compound word are provided. Further, the present invention is characterized in that the storage unit further includes a usage frequency information storage unit that stores usage frequency information indicating that the connection strength between the words exceeds a certain value. Further, the present invention is characterized in that the character string analysis unit converts the word into a compound word as a compound word based on the use frequency information.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of a character string search device according to an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a character string search device according to one embodiment of the present invention.

As shown in FIG. 1, a character string search device according to an embodiment of the present invention includes an input device 10 having a keyboard, a pen, and the like, and a character string analyzing unit 20 for analyzing input characters obtained from the input device 10. And a storage unit 3 for storing word information
0, a search processing unit 40 for extracting a document from the analysis result character string output from the character string analysis unit 20, a document database 50, and an output device 60 for outputting the result.

The storage unit 30 includes a word dictionary 32, an unnecessary word dictionary 34, a connection information storage unit 36, and a use frequency information storage unit 38. The word dictionary 32 stores various words and their parts of speech information in association with each other. Further, the connection information storage unit 36 stores connection information that is paired with the word dictionary 32 and quantifies the strength of connection between words and parts of speech. Further, the unnecessary word dictionary 34 stores word information that is not desirable to be matched as a search result.

The usage frequency information storage unit 38 stores usage frequency information indicating that the connection strength between parts of speech exceeds a certain value. This use frequency information is used as additional information of the connection information, and when words having the same connection strength appear at the next analysis, the information is analyzed with reference to this information.

The character analysis unit 20 determines a character string delimiter position of the character string input from the input device 10 in accordance with the word dictionary 32 and the unnecessary word dictionary 34.
The connection strength of each word is checked from the connection information storage unit 36, and a word having a certain degree of connection strength between words is extracted as a compound word. If the connection strength is less than a certain level, it is extracted using only that one word. Further, the use frequency information is added and stored in the use frequency information storage unit 38 for a word whose connection strength once exceeds a certain value and is regarded as a compound word.

The search processing unit 40 includes the character string analysis unit 20
Is input, and the document database 50 is searched for a matching document.
The character analysis unit 20 further separates and searches the compound words obtained from the character string analysis unit 20. This process is repeated until one word is finally obtained. Then, information of all the finally matched documents is output by the output device 60 such as a display, and a search result is obtained.

Next, the operation of the character string search apparatus according to one embodiment of the present invention having the above configuration will be described in detail. FIG. 2 is a flowchart showing the operation of the character string search device according to one embodiment of the present invention.

First, when a character string serving as a keyword is input from the input device 10 (step SA1), the input character string is passed to the character string analysis unit 20. Character analysis unit 20
Analyzes whether the passed character string is included in the word dictionary 32 or the unnecessary word dictionary 34 (step SA)
2). As a result of this analysis, there are three types of words: words that exist in the word dictionary 32, words that exist in the unnecessary word dictionary 34, and words that do not exist in any of them.

The character string analysis unit 20 determines that the word is a word dictionary 32
If the word exists, the process is performed on the word. That is, words before and after this word are stored in the unnecessary word dictionary 34.
It is stored whether or not it exists. If both words before and after the word are words that are not in the unnecessary word dictionary, the process ends.

Next, the character string analyzing section 20 executes the processing in the word dictionary 32
The connection strength between the words is checked from the connection information stored in the connection information storage unit 36 for a word in which a plurality of words exist in the same manner (step SA4). that time,
The values of the connection strengths are summed, and those having a certain value or more are extracted as compound words.

By the above processing, the character string input from the input device 10 is divided into the following five. (1) Compound word obtained from connection strength (2) One word whose connection strength is weak (3) One word whose preceding and following words are not in the word dictionary (4) Word obtained from unnecessary word dictionary (5) Words that do not exist in the dictionary

From above, the above (1), (2), (3),
The word (5) is extracted and passed to the search processing unit 40. When the word is passed, the search processing unit 40 performs a search process on the document database 50 (step SA).
4) Acquire matched document information.

With respect to the word (1), a search process is performed after the analysis process (step SA2 in FIG. 2) is performed again. This process is repeated until the word of (1) is exhausted. The search processing unit 40 performs a search process by merging all the results (step SA3), and outputs the search results to the output device 60. When the search result is output from the search processing unit 40, the output device 60 outputs the search result (step SA5).

Next, an embodiment of the present invention will be described in more detail using a specific example. FIG. 3 is a diagram for describing a specific example of one embodiment of the present invention in detail. Now, it is assumed that the character string "about the yen exchange rate in the Tokyo foreign exchange market" (FIG. 3B) is input to the input device 10 in FIG. This character string is passed from the input device 10 to the character string processing unit 20.

The character analyzer 20 divides the received character string based on the contents of the word dictionary 32 and the unnecessary word dictionary 34. It is assumed that the character strings are separated as shown in FIG. In other words, the string is the words "Tokyo""Foreign"
"Exchange""Market""NO""Yen""NO""Market""NI"
It is assumed that it is separated as "about". When the above processing is completed, each of the words separated next is classified into three types: those in the word dictionary 32, those in the unnecessary word dictionary 34, and those not in any of them. In the example of FIG. 3C, the word dictionary 32
Those in the unnecessary word dictionary 34 are denoted by the symbol "1", and those in the unnecessary word dictionary 34 are denoted by the symbol "2".

Then, referring to the classification of the words before and after each further divided word, "Tokyo", "Foreign", "Exchange", "Market"
Is determined to be likely to be connected from the connection information stored in the connection information storage unit 36, and is extracted as a compound word (FIG. 3 (d). The words "circle" and "quote" are not necessary). "No" and "No" in the word dictionary 34
Since it is surrounded by “ni”, “circle” and “market” are one word (FIG. 3E).

As a result of the above analysis by the character string analysis unit 20, the words shown in FIG. 3F are passed to the search processing unit 40, and the search processing unit 40 stores the document database 50 based on these words. Search for. Upon completion of the search processing, the search processing unit 40 extracts the obtained document as a first document group.

Next, the character string analyzer 20 analyzes the combined compound word (in this case, "Tokyo Foreign Exchange Market") again. This analysis is performed by the connection information storage unit 36 stored in the word dictionary 32 and the connection information storage unit 36 as described above.
To set the delimiter of the compound word. In this case, "Tokyo""Foreignexchange""Foreign exchange market" (Fig. 3
It is assumed that information that can be cut as shown in (e)) is stored. This process is obtained by registering the connection between words in advance.

Then, based on the result, the search processing unit 4
0 performs the search again to rank. As a result, the obtained document group is referred to as a second document group, and the one searched by the words shown in FIG. 3F is referred to as a third document group. At this time, the word already searched in the previous document group (the word “Tokyo” in FIG. 3E) is removed. By repeating the above processing until all words are separated, a third, fourth,..., N-th document group is obtained. The output device 60 outputs the finally obtained first to n-th document groups as search results in order.

Next, another embodiment of the present invention will be described. This embodiment is different from the above-described embodiment in that
The point is that not only the connection information stored in the connection information storage unit 36 but also the usage frequency information stored in the usage frequency information storage unit 38 is used as the information related to the word dictionary 32.

When a character string is input from the input device 10,
The character string analysis unit 20 classifies the character string input from the input device 10 into a compound word, a word, and the other based on the connection information of the words stored in the connection information storage unit 36. At this time, the value of the use frequency information for the word obtained as the compound word is increased. On the other hand, if no compound word is found in the document, the value of the usage frequency information is reduced. This usage frequency information is used when a compound word of three or more words appears in the next search, or when the compound word does not exist in the document.

For example, suppose that a compound word consisting of a certain three words is obtained from the input, and each word is represented by "1" and "2".
Assume “3”. At this time, if the word consisting of “123” is not found in the document, “1” is used in the above-described embodiment.
The search is divided again into “2” and “3”, and the search is performed again. Therefore, documents containing only “1”, “2”, or “3” match. Therefore, if the connection of “12” is more frequent than the connection of “23” based on the use frequency information obtained in advance, “12” and “3” are searched for, and “1”, “2”, “2” It can be processed to search for a document that is closer to the input than one containing only one “3”.

Next, the operation of this embodiment will be described as a more specific example. FIG. 4 is a diagram for describing a specific example of another embodiment of the present invention in detail. First, as shown in FIG. 4B, consider a case where a character string “about the yen price in the Tokyo foreign exchange market” is input from the input device 10. When the character string is input to the character string analysis unit 20, the character string analysis unit 20 converts the received character string into a word dictionary 32.
Then, based on the unnecessary word dictionary 34, a process of separating the character string is performed. As shown in FIG. 4 (c), the character string is composed of the words "Tokyo", "Foreign", "Exchange", "Market", and "No".
It is divided into "yen", "no", "market", "ni" and "about".

Next, each of the separated words is stored in the word dictionary 32.
In the unnecessary word dictionary 34 and those in neither of them. Then, looking at the words further divided, it is determined from the connection information stored in the connection information storage unit 36 that the words “Tokyo”, “foreign”, “exchange”, and “market” are likely to be connected. Pull out (FIG. 4D).

Further, since the words "yen" and "quote" are surrounded by those in the unnecessary word dictionary 34, that is, "no", "no" and "ni", "yen" and "quote" are represented by one word. (Fig. 4
(D)). Therefore, as a result of the analysis, the words shown in FIG. These pieces of word information are passed to the search processing unit 40.

First, "Tokyo Foreign Exchange Market" appears as a compound word. For each of the words "Tokyo", "Foreign", "Foreign Exchange", and "Market", refer to the usage frequency information stored in the usage frequency information storage unit 38. I do. Now, it is assumed that the input "exchange market" has appeared before, and the use frequency is higher than that between "foreign countries" and "exchange". In that case, as a result of the re-analysis, search for "Exchange market" as level 1 and "Tokyo" as level 2
Search for "forex". As a result, it is expected that a document closer to the input will be found than "Tokyo", "Foreign exchange", and "Foreign exchange market" as a group of documents of the same level.

FIG. 5 is a diagram showing an example of documents having different ranks among the obtained documents. The document shown in FIG. 5A has a high rank, and the document shown in FIG. 5B has a low rank.

If the words of levels 1 to 3 do not exist, all the words shown in FIG. 4C are searched. When a document is found by the search processing unit 40, a character string is output from the output device 60. In this embodiment, the maximum break has been reached in three stages,
In the case of other character strings, this is repeated many times and processed up to the maximum break. In other words, a search is performed every time the delimitation of the character string is repeated, and the matched documents are arranged in order. As a result, a document including a character string closer to the character string input from the input device 10 can be output from the output device 60.

[0035]

As described above, according to the present invention, it is possible to find a document that is closer to the content of the input character string, instead of searching for a document that simply contains the input search character. There is an effect that can be. Therefore, there is an effect that the trouble of re-inputting, further inputting from the result, narrowing down the search, and the like can be omitted.
Further, by reducing the input time and the access time, there is an effect that the content of the acquired document can be referred to more.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a character string search device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an operation of the character string search device according to one embodiment of the present invention.

FIG. 3 is a diagram for explaining a specific example of an embodiment of the present invention in detail.

FIG. 4 is a diagram for explaining a specific example of another embodiment of the present invention in detail.

FIG. 5 is a diagram showing an example of documents having different ranks among the obtained documents.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Input device 20 Character string analysis part 30 Storage part 32 Word dictionary 34 Unnecessary word dictionary 36 Connection information storage part 38 Usage frequency information storage part 40 Search processing part 50 Document database

Claims (3)

[Claims] 1. An input device for inputting a character string to be searched, a word dictionary storing words, an unnecessary word dictionary storing word information that is undesirable as a search result, and a connection between words. A storage unit having a connection information storage unit that stores connection information indicating strength; and a character string input from the input device is divided into words based on the word information and the unnecessary word information, and based on the connection information. A character string analysis unit that uses the word as a compound word, and a search processing unit that searches the document database for the compound word. 2. The storage unit according to claim 1, wherein the storage unit further includes a usage frequency information storage unit that stores usage frequency information indicating that the connection strength between the words exceeds a certain value. Character string search device. 3. The character string search device according to claim 2, wherein the character string analysis unit converts the word into a compound word as a compound word based on the use frequency information.