JP2003186771A - Url designating apparatus and recording medium storing url designating program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a URL designating apparatus and a recording medium for storing URL designating programs, which allow users to connect to a target site only by inputting natural language. <P>SOLUTION: The URL designating apparatus 1 has a natural language input part 10 to input natural language, a browser part 11, which connects to a created URL and reads data from the URL site to display it, and a natural language analysis part 12, which retrieves keywords from inputted natural language. In addition, the URL designating apparatus 1 also has a database 3, a learning function part 13, which creates a target URL using the keywords retrieved based on accumulation information stored in a personal database 14 and performs learning to rearrange words and phrases that can be a URL according to a user's taste, and the personal database 14, which accumulates URL corresponding to keywords. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a URL designating device and a recording medium recording a URL designating program that enables a user to connect to a target site without having to make a troublesome URL (Uniform Resource Locators) input. .

[0002]

2. Description of the Related Art As the Internet is developing, T
There are many cases of displaying and exchanging a homepage URL and an inquiry e-mail address on V, radio, advertisements, and the like. Generally, the URL is a long alphanumeric character string, and the e-mail address is not so long as the URL, but it is still long and cannot be remembered only by hearing or seeing it once. WEB to solve these
The search services are well developed above.

In the case of making inquiries and requesting retrieval by e-mail as well, there are already existing systems for automatically responding to them and prompting the recipient for an answer.

Also, due to the progress of information technology and the spread of the Internet, an environment has been established in which end users can easily acquire a wide variety of computerized information. However, the flood of large amounts of computerized information far exceeds the amount that can be managed by humans, and it is difficult to find the desired information. In response to such a situation, a system for selecting information by a computer has been developed and put into practical use. For example,
Examples include a search engine based on keyword matching, information filtering that selects information according to the interest of the user, and information navigation using information filtering.

There are the following URL search devices for simplifying browsing on the Internet. Japanese Unexamined Patent Publication No. 2000-339333 discloses a search support technique using natural language when using the Internet. This technology analyzes the natural language search phrase input from the browser by an external natural language analysis unit,
From this analysis result, the corresponding URL from the external database
Is what you get. However, this technology is just a website search support function on the Internet, and the connection to the target site is HTTP (Hypertext Transf
er Protocol) and FTP (File Transfer Protocol) must be taken into consideration when constructing a link to an external database, which is troublesome. Also,
Natural language processing specialized for each user such as "I want to connect to a company's file server" is not possible, and accurate search is not always possible.

Further, Japanese Patent Laid-Open No. 11-88407 discloses a URL designation technique for simplifying complicated URL designation. This technology defines a search term described in a familiar language called a second address, and has a memory device that associates the second address with a normal URL, thereby simplifying the user's URL input. . However, with this technology, it is not possible to handle inputs other than the second address stored in advance in memory, and when such an input occurs, the processing is left to the site search by the "search engine". Become. Therefore, it is difficult to flexibly respond to the user's input. Also, if the target site is structured hierarchically,
It is necessary to define the second address in all the layers, which seems to increase the complexity.

[0007]

However, such a conventional search support device has the following problem to be solved. In browsing on the Internet, specifying a URL to reach a target site is very complicated, and it is difficult for a general user to understand. In view of such a situation, there is a technique as described above. However, in the device described in Japanese Patent Laid-Open No. 2000-339333, it is necessary to prepare a database in consideration of the connection protocol, and there is a problem that natural language processing specialized for each user becomes difficult. There is. Further, in the device described in Japanese Patent Laid-Open No. 11-88407, it is necessary to prepare a second address for each URL, and when the target site is hierarchical, the work related to the association becomes very complicated. There is a problem such as becoming. Whichever method is used, the user is required to make a setting for a specific server, display a search site to obtain desired information, and specify a complicated URL.

The present invention has been made in view of the above problems, and it is possible to connect with a target protocol from a target site simply by inputting a natural language into a browser, and a natural language specialized for each user is provided. It is an object to provide a URL designating device and a recording medium recording a URL designating program, which enables the construction of a language analysis environment and can reach a target site flexibly corresponding to a natural language input by a user. To do.

[0009]

A URL designating apparatus of the present invention includes a natural language input means for inputting a natural language, a database for accumulating a URL corresponding to a keyword, and a keyword based on the input natural language. A keyword extracting means for extracting a phrase, a URL creating means for creating a target URL from the extracted keywords using the accumulated information in the database, and the created U
And a connection means for connecting to the RL site.

Further, the URL creating means is characterized in that the URL is created by reading the URL corresponding to the phrase extracted by the keyword extracting means from the database. In addition, more preferably, the URL creating means uses a plurality of URs from the database.
When L is read, the URL is given in a predetermined priority order.
May determine the presentation order of the URL
The creating means may raise the priority of the URL regarding the verb portion of the extracted phrase.

Further, the present invention comprises a preference information storage means for storing the preference information of the user, wherein the URL creating means creates the URL based on the user's preference information.
By rearranging according to the user's preference, it becomes possible to automatically generate the URL to the site indicated in the natural language input according to the user's preference.

Further, the URL creating means can edit according to the user's preference by switching the URL rearranging method based on the user's preference information, and reachability to the target site is possible. Can be increased. The URL designating apparatus of the present invention thus configured can reach a target site by inputting a commonly used natural language.

Further, according to the present invention, the keyword extracting means stores a word dictionary that stores a pair of a keyword phrase and a meaning vector representing the meaning thereof, and a phrase extracted from the input natural language. The sentence analysis means for meaning interpretation using the words registered in, the meaning vector calculation means for calculating the meaning vector of the meaning interpreted words, the words and meaning vectors calculated by the meaning vector calculation means By providing a control means for controlling the registration in the word dictionary, the keyword of the phrase included in the input natural language is automatically detected, and the URL to the site according to the user's intention based on the meaning vector of the keyword. Can be automatically generated.

Further, since the database is a personal database for accumulating URLs corresponding to keywords according to the taste of the user, it is possible to appropriately perform processing for natural language input according to the taste of the individual. Become. Further, the database is a URL corresponding to general / general-purpose keywords connected to an external network.
Since it is a public database for accumulating, it becomes possible to accurately process the input of general names and the like. In addition, more preferably, the connection unit may include a browser that reads and displays data of the connected URL site.

Further, according to the present invention, a natural language input means for inputting a natural language, a database for accumulating URLs corresponding to the keywords, and a keyword extracting means for extracting a word or phrase serving as a keyword based on the input natural language. And a URL creating unit that creates a target URL from the extracted keywords using the stored information in the database, and a connecting unit that connects to the created URL site. A computer-readable recording medium having a program recorded therein.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The preferred embodiments of the URL designating apparatus of the present invention will be described in detail below with reference to the accompanying drawings. First Embodiment FIG. 1 is a diagram showing a configuration of a URL designation device according to a first embodiment of the present invention. This URL designating device is an information processing device such as a personal computer or a PDA (Personal D
It can be applied to all information devices such as mobile information terminals such as igital Assistants).

In FIG. 1, reference numeral 1 denotes a UR which is troublesome to the user.
A URL designating device by natural language analysis that enables connection to a target site by inputting a commonly used natural language without inputting L, 2 is a network connectable to the URL designating device 1, Reference numeral 3 is a public database that is connected to the network 2 and is connected to an external network and enables general / general-purpose keyword conversion.

The URL designating device 1 is a natural language input section 10 for inputting a natural language which is commonly used in daily life.
(Natural language input means), a browser section 11 (connecting means) for connecting to the created URL and displaying a page, a natural language analyzing section 12 (keyword extracting means) for extracting a keyword phrase from the input natural language, Database 3,
A learning function unit 13 (URL creating means) for creating a target URL from the extracted keywords using the accumulated information of 14 and performing learning for rearranging words or phrases that can become URLs according to the user's preference or the like, And a personal database 14 for converting the extracted keywords into phrases that can become URLs.

Natural language analysis section 12 and learning function section 1
3, the personal database 14 and the public database 3 connected through the network 2 are URs as a whole.
The L creating unit 15 is configured. The URL creating unit 15 guesses and creates a target URL from the input natural language.

The browser unit 11 reads and displays the data of the connected URL site, for example, WWW (Worl).
d Wide Web) browser. The natural language analysis unit 12
It extracts keywords that are keywords based on the input natural language, and the analysis method can be rewritten according to the user's taste.

The personal database 14 is a personal database for accumulating URLs corresponding to keywords according to the taste of the user. In addition, personal database 14
Has a function as a user preference information storage unit that stores information that reflects the user's preference, converts the extracted keyword into a phrase that can be a URL, and performs keyword conversion according to the user's preference. , Conversion keywords and the like can be rewritten according to user preference. The public database 3 is a UR corresponding to general / general-purpose keywords connected to an external network.
It is a database that stores L.

The URL designating device 1 of the present embodiment can send a message to another device through the network 2 and receive a message from another device through the network 2. Here, the URL designation device 1 may be, specifically, an information device such as a personal computer, a mobile terminal, and a mobile phone that can be connected to the server by wire or wirelessly, and the public database 3 is a commercial content information. May be a server that delivers the. Further, the URL designating device 1 includes a data transmitting / receiving unit (not shown), and the interface may be a wireless interface or a wired interface connected to a cable.

The network 2 includes the URL designating device 1
Are connected to any number of units. Further, the number of terminals constituting the network, the configuration of the network, the form of the data transmission path, wired or wireless, etc. are not limited.
The network 2 is a network including a mobile communication network, a public telephone network, a LAN, the Internet, and the like, which connects a plurality of devices including the URL designating device 1. The network 2 includes a network type and a protocol such as a wired system or a wireless system. The type does not matter. The Internet is an analog and digital network that is spread all over the world, and sends and receives mail with other terminals via this network. Network 2
FTTH (Fiber To The Hom
e), HFC (Hybrid Fiber Coax),
And ADSL (Asymmetric Digital Subscriber Line)
A large capacity line such as is available.

FIG. 2 is a diagram showing a hardware configuration of the URL designating device 1. In FIG. 2, the URL designating device 1
Is an information processing device such as a personal computer, and a display device 13 including a CPU 110 (keyword extracting means, URL creating means) for controlling the entire device, a memory 120, an LCD display, drivers, and the like.
0, an input device (natural language input means, preference information input means) including a pointing device such as a keyboard and a mouse for inputting data and instructing various functions,
And an external interface (IF) 140 which is a light receiving unit of a remote controller by infrared communication, an HDD for storing various information
150 (personal database, user preference information storage means), recording medium driving device 160, network interface 170 (connecting means) including a modem port unit for connecting to the network 2 through a telephone line, and a bus connecting these to each other. 180 is included.

The URL designating device 1 communicates using the network interface 170 to connect to the network 2 such as the Internet. The CPU 110 is a memory 1 that stores a control program and data used for calculation.
20 is used to execute various processes of the information terminal.

In particular, the CPU 110 causes the URL designating device 1 based on an instruction from the OS program or the user program.
Controlling the operation of each of the above elements to control the entire apparatus including execution of application processing such as URL designation processing, data reception processing, data analysis processing,
Performs user interface processing, etc. Also, CPU1
10 determines whether or not the detected word is a word already registered in the word dictionary, and when the word already registered in the word dictionary is detected, the calculated meaning vector of the word and the word The meaning vector of the word registered in the dictionary is calculated according to a predetermined calculation formula, and the meaning vector of the word registered in the word dictionary is updated by the calculated meaning vector. Note that extraction of related words and phrases using the meaning vector and URL generation by the extracted words and phrases will be described later in the second embodiment.

The memory 120 is composed of a ROM, a RAM and the like, and the ROM is a read-only semiconductor memory for storing fixed data such as programs and communication control data required when the CPU 110 operates. Further, the ROM stores control programs and device-specific information. RAM
Is used as a so-called working memory for temporarily storing data relating to character display, data used for calculation, calculation results, and the like. A part of the RAM is an electrically rewritable nonvolatile memory such as an EEPROM (elec
trically erasable programmable ROM), EE
By changing the program written in PROM,
Various specifications of the URL designating device 1 can be changed. That is, recently, in order to avoid the time loss of changing the mask ROM for each debugging of system development,
The program ROM is a non-volatile memory such as EPRO.
With M and EEPROM, the program development and modification time is greatly shortened. Further, by downloading the program and rewriting the program contents of the EEPROM, it becomes possible to easily upgrade the function or change the function.

The memory 120 is loaded with the OS program and the user program, and temporarily stores the data received from the network and temporarily records the received data. As the memory 120, a semiconductor memory such as SDRAM that can be accessed at high speed is used.

The display device 130 is an LCD display or a CRT display connected to the URL designating device 1. The user looks at the display screen of the display device 130 and
Various kinds of information can be input by using an input device such as a keyboard or a mouse. Further, when an external display is connected to the URL designating device 1, an image can be displayed on this external display.

The external IF 140 is an input device such as a mouse or a keyboard connected to the URL designating device 1, an infrared ray (I
R) is an interface for connecting an IR receiving unit that performs communication in R), an external recording device such as the HDD 150, and various devices. The user can use these input devices to input necessary instructions on the instruction screen or the like displayed on the display device 130.

The external IF 140 is specifically a PCMC.
IA (Personal Computer Memory Card Interface Asso
The card connection unit is composed of an insertion slot and a connector for connecting a PC card interface unit, SD (Secure Digital) memory card, CompactFlash (CF) card, etc., which are defined as ciation standards. In addition, as a communication means of the URL designating device 1,
In addition to wired communication, Bluetooth, wireless LAN
Other wireless communication technologies, such as

The HDD 150 is a large capacity hard disk capable of recording a large amount of data at high speed. Further, a device that can read another data even while writing the data is used.

The recording medium driving device 160 is used to input and output the removable recording medium 161. The recording medium 161 stores a user program or the like for realizing the present URL designating method. In that case, the program is read from the recording medium driving device 160 and stored in the memory 12.
Loaded to zero. In addition, user preference data, a keyword dictionary, etc. may be recorded.

The HDD 150 is an external storage device for recording the data selected by the CPU 110. In addition to a disk device such as a hard disk drive (HDD), the HDD 150 is, for example, an FD, a CD-ROM, a CD-R / RW, a DVD.
(Digital Versatile Disc), disk devices such as hard disk drives (HDD) and MO (Magneto Optical), PC cards that are SRAM (StaticRAM) cards that retain information written by power backup, and flash memory that does not require power backup Compact flash (CF) (registered trademark), smart media (registered trademark), and memory stick (registered trademark).

The network interface 170 is U
It is an interface for connecting the RL designating device 1 to the network 2, and is a peripheral device that transmits and receives digital data through a communication line such as a public telephone network. Depending on the form of connection, a satellite receiver, a modem or a router may be required, but in that case, these are included.

The bus 180 is a common transmission path for transferring various kinds of data transmitted and received among the above-mentioned constituent elements at high speed. Hereinafter, the operation of the URL designating device configured as described above will be described.

The URL designating apparatus 1 enables the user to reach the target site by natural language input in order to reduce the troublesome inputting of the URL required for Internet browsing. The natural language input to the natural language input unit 10 is sent to the URL creating unit 15, converted into a candidate URL by the URL creating unit 15, and then passed to the browser unit 11. The operation of the URL creating unit 15 is as follows.

The natural language input by the user is sent to the natural language analysis unit 12. The natural language analysis unit 12 extracts a keyword phrase from the input natural language and sends the extracted keyword to the personal database 14 and the public database 3. The personal database 14 and the public database 3 collate the acquired keywords with their own databases, and perform conversion work into words that can be URLs. The words and phrases that have undergone this conversion work are sent to the learning function unit 13, and the words and phrases are rearranged in accordance with individual tastes to create a URL. The natural language analysis unit 12, the learning function unit 13, the personal database 14, and the like can be rewritten by each user, and can reach the site more suited to the user's taste.

For example, by associating "company" with "www.sharp.co.jp" in the personal database 14, the input in natural language becomes more suitable for the user. An example of this series of operations will be described with a specific example based on FIG.

FIG. 3 is a flow chart showing the operation of each part of the URL designating device 1 by natural language analysis. This flow is executed by the CPU 110 as a URL designing related process. In the figure, S indicates each step of the flow. A is a natural language character string, B is a keyword phrase,
C is a phrase that can be used in a URL, D is a phrase that can be used in a URL, and E is a URL that is a candidate.

When the present URL designating program is started according to the user's instruction, first, in step S1, the user inputs a natural language commonly used in daily life through the natural language input unit 10. As a natural language character string A,
Take the case where "I want to see new product information of the company" is entered. The input natural language character string A is sent to the processing by the natural language analysis unit 12.

In step S2, a word or phrase serving as a keyword is extracted from the input natural language character string A. In this case, the natural language analysis unit 12 causes the keyword B to be a keyword B “company / new product information / want to see” from this character string A “want to see new product information of the company” (/ indicates a keyword delimiter). And sends it to the personal database 14 and the public database 3. In addition, in the present embodiment, the natural language analysis unit 12 extracts a word or phrase that is a keyword registered in advance from the natural language, and does not extract a word or phrase other than the keyword. An example of extracting a related phrase from the input natural language character string using a semantic vector will be described later with reference to the second embodiment.

In the present embodiment, the URL designating device 1 is provided with the personal database 14 and is connected to the public database 3 via the network 2. However, the personal database 14 or the public database 14 is used. The configuration may be such that only one of the databases 3 can be accessed. In FIG. 3, an example using the personal database 14 and the public database 3 will be described. In this case, the usage conditions of the databases 14 and 3 may be different. For example, the use of the personal database 14 is prioritized, and when there is no corresponding data in the personal database 14, the personal database 14 is connected to the public database 3 via the network 2, or the personal database 14 and the public database 3 are used. A mode in which the URL obtained using the personal database 14 is presented higher than the URL obtained using the public database 3 can be considered. In the above case, after step S2, a step of confirming the usage conditions of the database and the URL presentation method with the user may be provided.

Returning to the flow, the personal database 14
The public database 3 and the public database 3 are based on their own databases, and the keyword B is "company / no /".
New product information / want to see "is converted into words C and D that can be used in the URL.

In the URL generation process using the personal database 14 in step S3, for example, the words and phrases B "company", "no", "new product information", and "want to see" can be used in the URL, respectively. www.shar
p. co. jp or sharp-world. com etc. "" / "" whats new or new product
It is converted into “cts etc.” “http: /// or rtsp: /// etc”.

Further, in the URL generating process using the public database 3 in step S4, for example, the keyword B "company""no""new product information""want to see"
The phrase D “www.ka” that can be used in the URL
isya. co. jp or company. com etc. "" / "" information or whats
-New, etc. "" http: ///, mms: // or h
"https: // etc.".

These words C and D are sent to the learning function processing in step S5, where they are rearranged based on personal tastes, and the browser section 11 in step S6.
Then, browsing is performed in order from the URL having the highest probability of becoming the target site. For example, in the above example, "http: // www
w. sharp. co. jp / whatsnew / ”
"Http: //www.sharp-world.c
om / whatsnew / ”“ http: // kais
ha. co. jp / ”...

In the browser section 11, the acquired URLs are sequentially connected, and the acquired pages are displayed on the display device 130 as a browser with a tag attached.

In addition, the natural language analysis processing in step S2 and the U using the personal database 14 in step S3 are performed.
The RL generation process and the learning function process of step S5 can be edited according to the user's preference, and each conversion process for the input natural language can be adapted to the user's preference. , It is possible to increase the reachability to the target site. These will be described later with reference to FIG.

By executing the above flow, specifically, the following URL generation is realized. (1) When the user inputs "I want to download from abc corporation" -abc corporation → abc. co. jp, abc. co
m, www. abc. co. jp, ftp. abc. c
o. jp ・ I want to download → Extract words and phrases such as ftp: ///, http: ///, etc. and perform URL conversion of the words, ftp: /// abc. co. jp / ftp: /// abc. com / ftp: // ftp. abc. co. jp / http: // www. abc. co. jp / inde
x. A URL such as html is generated.

Here, in the word rearrangement procedure in the learning function process, if the verb part (in this example, "I want to download") is weighted, UR
In L generation, www. abc. co. ft rather than jp
p. abc. co. You can raise the priority of jp,
It enables more accurate conversion according to the user's intention.

(2) When the user inputs "I want to watch the video of movie 1 recorded in the home video server", the URL of the home video server (video.foo.bar) is stored in the personal database 14 in advance. Is registered.・ Home video server → video. foo. bar ・ Movie 1 → movie1. asf, video1. as
f, ... ・ I want to watch the video → mms: //, rtsp: //, h
.. and the like are extracted and the URL is converted into the URL. mms: // video. foo. bar / movie
1. asf rtsp: // video. foo. bar / movi
e1. asf mms: // video. foo. bar / movie
1. A URL such as asf is generated.

(3) When the generated URL does not exist For example, as the URL, http: // www. foo. bar / home / i
ndex. If html is generated and the connection fails because it does not exist, a message prompting rewriting of the input phrase is displayed. Also, if the connection server permits, http: // www. foo. ba
The file name existing in r / home / is displayed.

(4) When the URL cannot be generated because a word or phrase not registered in the personal database 14 or the public database 3 is input, a message indicating that the URL cannot be generated or the input word or phrase is urged to be rewritten is displayed. indicate. As will be described later in the second embodiment, if a related word is extracted using a semantic vector, the possibility of URL generation can be significantly increased. In this case, when the URL generation by the related phrase using the meaning vector also fails, a message or the like for prompting rewriting of the input phrase is displayed.

FIG. 4 is a functional block diagram showing an example of natural language processing editing in the natural language analysis unit 12.

First, the natural language analysis unit 12 may be edited in order to perform an accurate analysis according to the dialect in each region or the habit of document input for each individual. The natural language analysis unit 12 is equipped with a natural language analysis engine 21, which is the center of analysis. Natural language analysis engine 2
When 1 is largely edited (for example, the Kanto valve analysis engine is replaced with the Kansai valve conversion engine), each natural language analysis engine 21 is replaced with a new natural language analysis engine 22. This work is performed by downloading the natural language analysis engine using the Internet network or by delivering the data from each company. On the other hand, if there is a slight change in the analysis such as the fluctuation of the ending, the editing work can be completed by changing the setting inside the analysis engine.

Next, the editing of the personal database 14 will be described.

FIG. 5 is a functional block diagram showing an example of personal database editing in the natural language analysis unit 12.

It is considered that the personal database 14 is edited so that appropriate conversion can be performed according to a specific noun or expression of each individual. Therefore, the personal database 14 has a correspondence table 31 as shown in FIG. For example, in the correspondence table 31 of FIG. 5, the user selects "Mr. Yoshida" as "www.yoshida.co."
When it is desired to associate it with m ”, this information is written in the correspondence table 31. By adding the portion 32 added in this way, more accurate natural language processing can be performed.

For example, a database editing application for carrying out registration work such as the word registration function of the Japanese conversion engine is prepared for the personal database 14. Use this editing application to change "home" to "foo.ba
If you want to associate it with "r"
By inputting "foo.bar" as the corresponding phrase, registration is performed according to the format of the database.

The learning function unit 13 edits the addresses created from the natural language in order to display in order from the highest expected page probability.

FIG. 6 is a functional block diagram showing an example of learning function editing in the natural language analysis unit 12.

The learning function unit 13 forms an access list 41 of sites that are often connected or sites that cannot be connected, for example, from past Internet access history. Based on this information, the learning function unit 13 sequentially outputs the URLs that are considered to be the closest to those designated in the natural language. In the example of FIG. 6, “http: // www
w. sharp. co. jp / ”is sorted in ascending order.

As described above, the URL of this embodiment
The designation device 1 includes a natural language input unit 1 for inputting a natural language.
0, a browser unit 11 that connects to the created URL and reads and displays the data of the URL site, a natural language analysis unit 12 that extracts a keyword phrase from the input natural language, and information stored in the databases 3 and 14 A learning function unit 13 for creating a target URL from the extracted keywords and learning for rearranging words and phrases that can become URLs according to the user's preference, and a personal database for accumulating URLs corresponding to the keywords 14
Since it is configured with, it is possible to connect to the target site simply by inputting a commonly used natural language without the user having to input a URL that is bothersome or difficult to search. .

That is, in browsing on the Internet, specifying a URL to reach a target site is very complicated and is difficult for a general user to understand. However, the URL specifying device 1 inputs a natural language into a browser. By doing so, it is possible to connect with the target protocol from the target site. Further, since it is possible to build a natural language analysis environment specialized for each user, it is possible to achieve the target site flexibly corresponding to the words entered by the user.

Second Embodiment FIG. 7 is a diagram showing the structure of a URL designating apparatus according to a second embodiment of the present invention. In the description of this embodiment, the same components as those in FIG. 1 will be assigned the same reference numerals and overlapping description will be omitted.

In FIG. 7, the URL designating device 50 includes a natural language input section 10 for inputting a natural language and a created UR.
A browser unit 11 that connects to L and displays a page, a natural language analysis unit 51 (keyword extraction unit) that extracts a word or phrase that is a keyword from the input natural language and calculates a meaning vector of the word or phrase from the keyword included in the natural language , A target URL is created from the extracted keywords using the information stored in the databases 3 and 52, and U
The learning function unit 13 that performs learning for rearranging words that can be RL according to user's preference and the like, and a personal database 52 (semantic vector database) for converting extracted keywords into words that can be URLs Consists of

Natural language analysis section 51, learning function section 1
3, the personal database 52 and the public database 3 connected through the network 2 are URs as a whole.
The L creating unit 53 is configured. The URL creating unit 53 guesses and creates a target URL from the input natural language.

The natural language analysis unit 51 uses a word dictionary that stores a pair of a keyword phrase and a meaning vector that represents its meaning as a vector, and semantically interprets the phrase extracted from the input natural language to obtain the meaning vector. While calculating
It has a semantic vector calculation / registration function for controlling the registration of the calculated word and its meaning vector in the word dictionary.

The natural language analysis unit 51 extracts a word or phrase serving as a keyword based on the input natural language, and calculates a meaning vector of the word or phrase from the keyword included in the natural language, thereby extracting a wide range of related words or phrases. You can Further, as in the first embodiment, the analysis method can be rewritten according to the user's preference.

The personal database 52 is a personal database that stores URLs corresponding to keywords according to the user's taste. In addition, the personal database 52
Has a function as a user preference information storage unit that stores information that reflects the user's preference, converts the extracted keyword into a phrase that can be a URL, and performs keyword conversion according to the user's preference. , Conversion keywords and the like can be rewritten according to user preference.

Further, the personal database 52 has a function as a meaning vector database for accumulating meaning vector data, and stores a word dictionary for storing pairs of a keyword phrase and a meaning vector expressing its meaning in vector. .

FIG. 8 is a block diagram showing the structure of the natural language analysis section 51. In FIG. 8, a natural language analysis unit 51 extracts a keyword and a related phrase included in a natural language or the like using a semantic vector, and a data for extracting a text string from the data input by the natural language input unit 10. A decoding unit 61, a sentence analysis unit 62 (sentence analysis unit) that interprets the meaning of the extracted text string using the words registered in the word dictionary, and extracts words that are not registered in the word dictionary. A semantic vector calculation unit 63 that calculates the semantic vector of the extracted word
(Semantic vector calculation means), a similarity calculation unit 64 that calculates the similarity between the calculated phrase and the user preference information stored in the user preference information storage unit 66, and a word using the calculated phrase and its meaning vector as keywords. A determination unit 67 (control unit) that performs control for registering in the dictionary, a keyword registration unit 65 that includes a word dictionary that stores a pair of a character string of a word and a meaning vector that represents the meaning thereof, and a user's preference are reflected. User preference information recording unit 6 that stores information
It consists of 6.

The data decoding section 61 converts the input data into text data.

The sentence analysis unit 62 has a keyword registration unit 65.
Data decoding unit 6 based on the keywords registered in
The sentence of the text data obtained in 1 is analyzed.

The meaning vector calculation unit 63 is the sentence analysis unit 6
In the sentence analyzed in 2, the words that are not registered as keywords are extracted, and the meaning vector is given to the phrase.

The keyword registration unit 65 registers keywords necessary for analyzing the meaning of the received information. Specifically, the keyword registration unit 65 allocates a partial area of the personal database 52 as a meaning vector database and stores the word dictionary in the meaning vector database. The word dictionary may be stored in a storage medium physically or logically different from the personal database 52.

The user preference information recording section 66 stores or constructs or feeds back the user preference as a user profile based on the meaning content of the received information.

The similarity calculation unit 64 calculates the similarity between the word / phrase extracted by the meaning vector calculation unit 63 and the keyword stored in the user preference information recording unit 66 according to the user preference, and the similarity of the similarity is calculated. Sort in ascending or descending order.

The judging section 67 has a function as a control means, controls to register the calculated word and phrase and its meaning vector in the word dictionary as keywords, and determines the word or phrase based on the similarity calculation result by the similarity calculating section 64. Controls to determine the display order.

The operation of the URL designating apparatus configured as described above will be described below.

Since the overall operation of the URL designating device 50 is similar to that of the URL designating device 1 of the first embodiment, the description thereof will be omitted, and the natural language analysis operation using the semantic vector will be described in detail.

In the data decoding section 61, the natural language from the natural language input section 10 is input as data to the decoding section 61 and converted into text data by the data decoding section 61. The text data obtained by the data decoding unit 61 is sent to the sentence analysis unit 62, and the sentence analysis unit 62 analyzes the sentence. At the time of sentence analysis, analysis is performed based on the keywords registered in the keyword registration unit 65. In the sentence analyzed by the sentence analysis unit 62, words that are not registered as keywords are extracted by the meaning vector calculation unit 63, and a meaning vector is given to the phrase.

The similarity between the word / phrase extracted by the meaning vector calculation unit 63 and the keyword according to the user preference stored in the user preference information recording unit 66 is the similarity calculation unit 6
4 and are sorted in ascending or descending order of similarity.

In the judging section 67, in response to the user's operation,
The processing of correcting the range of words and selecting words to be registered is performed.

Next, a method for obtaining a meaning vector of a word or phrase processed by the sentence analysis unit 62 will be described.

The natural language input from the natural language input unit 10 is converted into text data by the data decoding unit 61 and sent to the sentence analysis unit 62. The contents are analyzed. Of the sentences analyzed by the sentence analysis unit 62, words not registered as keywords are extracted by the meaning vector calculation unit 63,
A semantic vector is attached to each word.

The semantic vector is one method of expressing the meaning of a word or a sentence by a numerical vector, and is an element (herein referred to as core or core word) that describes the characteristics of the search target.
Is a vector representation of the relationship with. For example, when 100 kinds of core words are prepared for semantic expression, "0" and "1" indicate that a certain word or sentence is "related" or "not" related to these 100 core words, It is used as a 100-dimensional vector to represent the meaning of words and sentences. The search results are ranked using both the distance of the generated meaning vector and the distance according to the word content rate, and the search results are presented in order of closeness. To be more specific, the basic words "travel" and "tour" are now 100
Core words (politics, planning, machines, movement, ...) are defined. "1" is set to those that match this core word, and "0" is set to those that do not.

[0089]   "Travel" (politics, planning, machinery, movement, mathematics, entertainment, experience, ...)             (0, 1, 0, 1, 0, 1, 1, 1, ...)   "Tour" (politics, planning, machinery, movement, mathematics, entertainment, experience ...)             (0, 1, 0, 1, 0, 1, 1, 1, ...)

As described above, the 100-dimensional meaning vector of "travel" and "tour" is [0,1,0,1,0,1,1, ...], and since the two vectors are almost the same, they have the same meaning. Judge as a word. Further, according to the following equations (1) to (3), the distance SVi between the user's request sentence and the search target semantic vector,
The total distance Si of the search results is obtained by using both the distances SWi depending on the word content rate, and the search results are ranked and displayed in order of closeness to meaning.

[0091]

[Equation 1]

For details of this technique, see Japanese Patent Laid-Open No.
No. 0-55372.

The method for obtaining the meaning vector of a word or phrase processed by the sentence analysis unit 62 is as follows. FIG. 3
This will be described with reference to the example of the flow. For example, in step S1 of FIG. 3, the user sets the natural language character string A as “a
When "I want to see new product information on bc sale" is input, the sentence analysis unit 62 of the natural language analysis unit 51 analyzes the sentence using the keywords registered in the keyword registration unit 65. In this example, this character string A "ab
c) I want to see new product information on sale ”is decomposed into a phrase B“ abc sale / of / new product information / want to see ”which can be a keyword, and sends it to the personal database 52.

The meaning vector calculation unit 63 uses these keywords to calculate the meaning vectors of these terms according to the equation (4).

[0095]

[Equation 2] However, ‖X‖ represents the normalized vector of X, n is the total number of keywords included in the sentence Xi is the meaning vector of each keyword

All the words and phrases contained in this sentence are set to the same value as the meaning vector of this sentence. However, here, if the keyword of the phrase “abc sale” is not registered in advance, in the first embodiment, the phrase other than the registered keyword cannot be extracted. URL generation is not possible. In this case, a message prompting you to rewrite the input phrase will be displayed. On the other hand, in the present embodiment, the meaning vector of “sales (sales company)” and the meaning vector of “company” are calculated by the above-described meaning vector calculation method.
Since the vectors are almost the same, it can be determined that the words have the same meaning or related phrases.

As described above, since the related words and phrases can be extracted from the input natural language character string by using the meaning vector, it is possible to further improve the arrival rate to the site desired by the user.

Further, in the present embodiment, in addition to the above-described semantic vector calculation processing, a user preference information recording section 66 and a similarity degree calculation section 64 are further provided, so that the similarity degree between the user preference information and each program information can be calculated. calculate. Here, the meaning vector of the user's preference information is the average value of the meaning vectors of the keywords selected by the user, that is, the above equation (4).
It is in the value calculated by. The similarity calculation unit 64 rearranges the program information in order according to the similarity between the program information and the user preference. The determination unit 67 registers the obtained related term as a keyword in the keyword registration unit 65. The word / phrase registered as a keyword is registered as a user preference, and will be used as a keyword when calculating the meaning vector of the user preference from the next time.

Returning to the flow of FIG. 3, in the personal database 52, based on its own database, the phrase B “abc company / of / new product information / want to see” can be used as the URL C. Convert to. This "abc company" is a natural language phrase "ab" entered by the user.
It is a keyword of a related term obtained by calculating a meaning vector from "c sale".

In the URL generating process using the personal database 52, the keyword B, "abc company", is used.
"No", "New product information", and "I want to see" are URLs
C that can be used for "www.abc.co.jp or abc-world.com etc.""/""wha
tsnew or newproducts etc. "" http
p: // or rtsp: // etc. ".

These words and phrases C are sent to the learning function process of step S5, where they are rearranged based on personal tastes and the like, and the browser section 11 is browsed in order from the URL having the highest probability of becoming the target site. It For example, in the above example, a URL that has a high probability of becoming a target site
In order from "http: //www.abc.co.j"
p / whatsnew / ”“ http: //www.a
bc-world. com / whatsnew / ”“ h
ttp: // abc-kaisha. co. jp / ”...
Be sorted into.

In the browser section 11, the acquired URLs are sequentially connected, and the acquired pages are displayed on the display device 130 as a browser with a tag attached.

The above is an example in the case where the user inputs "I want to see new product information of abc sales", but "abc
Similarly, when "I want to download new product information for sale" is input, abc Co., Ltd. (keyword of a related phrase obtained by using a meaning vector from "abc sale") → abc. c
o. jp, abc. com, www. abc. co. j
p, ftp. abc. co. jp ・ I want to download → Extract words and phrases such as ftp: ///, http: ///, etc. and perform URL conversion of the words, ftp: /// abc. co. jp / ftp: /// abc. com / ftp: // ftp. abc. co. jp / http: // www. abc. co. jp / inde
x. A URL such as html is generated.

Here, if the verb portion (in this example, "I want to download") is weighted, the UR
In L generation, www. abc. co. ft rather than jp
p. abc. co. You can raise the priority of jp,
It enables more accurate conversion according to the user's intention.

If the generation of the URL by the related phrase using the meaning vector also fails, it is preferable to display a message or the like for prompting the user to rewrite the input phrase.

As described above, the natural language analysis unit 51 of the URL designating apparatus 50 of the present embodiment registers the data decoding unit 61 for extracting the natural language text string and the extracted text string in the word dictionary. A sentence analysis unit 62 that extracts a word or phrase that is a word that is not registered in the word dictionary, a meaning vector calculation unit 63 that calculates a meaning vector of the extracted word, and a calculated word and phrase A determination unit 67 that performs control to register the meaning vector in the word dictionary as a keyword, and a keyword registration unit 65 that includes a word dictionary that stores a pair of a character string of a word and a meaning vector that represents the meaning thereof are provided. Since the keywords and related phrases included in the natural language input by the user are extracted using the vector, the related phrases according to the user's intention are extracted. Can be widely extracted, it is possible to improve the success rate to a site desired by the user.

For example, if the user inputs "I want to download tour information of Brazil" when the URL based on the keyword "tour" is not registered in the personal database 52 in advance, the first implementation is performed. In form, personal database 14 (FIG. 1)
There was a possibility that the URL could not be generated using (it is possible if the public database 3 is used). In the present embodiment, a URL based on the keyword “travel”, which is very closely related to “tour”, can be generated by the semantic vector, and the user can reach the desired site by this URL and download the Brazilian travel information. can do. In this way, a URL that is bothersome to the user
Also, it is possible to connect to a target site by simply inputting a commonly used natural language without inputting a URL that is troublesome or difficult to search.

Here, in the present embodiment, since a word and its meaning vector are automatically registered in the word dictionary as keywords, it is troublesome to manually input a part of speech and meaning, and the accuracy of extraction is low, so that incorrect cutout occurs. Without being performed, a phrase that is not registered in the dictionary can be detected accurately and efficiently and registered in the word dictionary.

Further, in the present embodiment, a user preference information recording section 66 for recording information reflecting the user's preference,
Since the similarity calculation unit 64 that calculates the degree of similarity between the words and phrases extracted by the meaning vector calculation unit 63 and the user preference obtained by the user preference information recording unit 66 is provided, in the order that reflects the user's preference, Words and keywords can be presented and you can quickly connect to the optimal URL.

Furthermore, by recording the software describing the procedure for automatically detecting and registering the above-mentioned words in a computer-readable recording medium, the above-mentioned functions can be realized on the computer anytime and anywhere.

In each of the above embodiments, the URL designation device connected to the network is applied to an information terminal such as a personal computer and a PDA, and the public database is applied to a server. However, this is an example. , Applicable to any device. Further, the communication method of the communication system is not limited. Further, it may be incorporated as a data communication function of a portable information terminal or a personal computer.

Further, in each of the above-mentioned embodiments, the name "URL designating device" is used, but this is for convenience of description, and may be a natural language search support device, an information presenting device, an address search system or the like. Of course.

Furthermore, the type of each circuit section, the type of memory, the keyword extraction method, the learning method, the display method of the display screen, the connection method, and the like that compose the URL device are not limited to the above-described embodiments.

The URL designating apparatus described above is also realized by a program for operating this processing apparatus.
This program is stored in a computer-readable recording medium. In the present invention, as the recording medium, the main memory itself may be a program medium, or a program reading device is provided as an external storage device, and the program medium can be read by inserting the recording medium therein. May be. In either case, the stored program is CP
The configuration may be such that U accesses and executes the program, or in any case, the program is read, and the read program is downloaded to a program storage area (not shown) and the program is executed. May be It is assumed that this download program is stored in the main body device in advance.

Here, the program medium is a recording medium which can be separated from the main body, and is, for example, a PC card (SRAM card), a tape system such as a magnetic tape or a cassette tape, and a floppy disk (registered trademark).
Disk such as hard disk and hard disk, CD-ROM / M
Disk systems for optical disks such as O / MD / DVD, card systems such as IC cards (including memory cards) / optical cards,
Alternatively, it may be a medium that fixedly carries a program, including a semiconductor memory such as a mask ROM, EPROM, EEPROM, and flash ROM.

Further, in the present embodiment, since the system configuration is such that a communication network including the Internet can be connected, it may be a medium that carries the program fluidly so as to download the program from the communication network. . When the program is downloaded from the communication network in this way, the program for downloading may be stored in the main body device in advance, or may be installed from another recording medium. Further, the content stored in the recording medium is not limited to the program and may be data.

[0117]

As described above in detail, according to the present invention, it is possible to connect to a target site by a target protocol simply by inputting a natural language. In addition, it is possible to construct a natural language analysis environment that is specialized for each user, and it is possible to achieve the destination site that flexibly responds to the words entered by the user.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a URL designation device according to a first embodiment of this invention.

FIG. 2 is a diagram showing a hardware configuration of a URL designation device according to the present embodiment.

FIG. 3 is a flowchart showing a URL designating operation by natural language analysis of the URL designating apparatus according to the present embodiment.

FIG. 4 is a functional block diagram showing an example of natural language processing editing in a natural language analysis unit of the URL designating device of the present embodiment.

FIG. 5 is a functional block diagram showing an example of personal database editing in the natural language analysis unit of the URL designating device according to the present embodiment.

FIG. 6 is a functional block diagram showing an example of learning function editing in the natural language analysis unit of the URL designating device according to the present embodiment.

FIG. 7 is a diagram showing a configuration of a URL designation device according to a second embodiment of this invention.

FIG. 8 is a block diagram showing a configuration of a natural language analysis unit of the URL designating device according to the present embodiment.

[Explanation of symbols]

1,50 URL designation device 2 network 3 public database 10 Natural language input section (natural language input means) 11 Browser unit (connection method) 12,51 Natural language analysis unit (keyword extraction means) 13 Learning function section (URL creation means) 14,52 personal database 15,53 URL creation section 21,22 Natural Language Analysis Engine 31 Correspondence table 32 Added part 41 Access List 51 Natural Language Analysis Department 61 Data decoding section 62 Text analysis unit (text analysis means) 63 Semantic Vector Calculation Unit (Semantic Vector Calculation Means) 64 similarity calculator 67 Judgment unit (control means) 65 keyword registration section 66 user preference information recording unit A natural language character string B Keywords and phrases Words and phrases that can be used for C URL Words and phrases that can be used for D URL E Complementary URL

Claims (11)

[Claims] 1. A natural language input means for inputting a natural language, a database for accumulating a URL corresponding to a keyword, a keyword extracting means for extracting a phrase serving as a keyword based on the input natural language, and the database. A URL designating device comprising: a URL creating means for creating a target URL from the extracted keyword using the stored information of 1. and a connecting means for connecting to the created URL site. 2. The URL designating apparatus according to claim 1, wherein the URL creating unit creates the URL by reading the URL corresponding to the word or phrase extracted by the keyword extracting unit from the database. 3. The URL designating apparatus according to claim 2, wherein the URL creating means determines a presentation order of URLs according to a predetermined priority order when a plurality of URLs are read from the database. 4. The URL designating apparatus according to claim 3, wherein the URL creating means raises the priority of the URL regarding the verb portion of the extracted word. 5. A preference information storage unit for storing the preference information of the user is provided, and the URL creating unit is based on the preference information of the user.
The URL designating apparatus according to claim 1, wherein the created URLs are rearranged in accordance with a user's preference. 6. The URL designating apparatus according to claim 5, wherein the URL creating means switches the URL rearranging method based on the preference information of the user. 7. The keyword extracting means further registers a word dictionary storing a pair of a keyword phrase and a meaning vector expressing its meaning in a vector and a phrase extracted from the input natural language in the word dictionary. Sentence analysis means for interpreting the meaning using the defined words, a meaning vector calculation means for calculating the meaning vector of the meaning-interpreted words and phrases, the word and the meaning vectors calculated by the meaning vector calculation means, and the word dictionary 2. The URL according to claim 1, further comprising: a control unit that performs control for registering in the URL.
Designated device. 8. The UR according to claim 1, wherein the database is a personal database that stores URLs corresponding to keywords according to user preferences.
L designated device. 9. The database is a U corresponding to general and general-purpose keywords connected to an external network.
The URL designation device according to claim 1, wherein the URL designation device is a public database for accumulating RL. 10. The connecting means is the connected UR.
The URL designating apparatus according to claim 1, further comprising a browser that reads and displays the data of the L site. 11. A natural language input means for inputting a natural language, a database for accumulating a URL corresponding to a keyword, a keyword extracting means for extracting a keyword phrase based on the inputted natural language, and the database. A program for functioning as a URL designating device having a URL creating means for creating a target URL from the extracted keyword using the accumulated information of the above and a connecting means for connecting to the created URL site is recorded. A computer-readable recording medium characterized by the above.