JPH02114367A - Text processing device and method - Google Patents

Abstract

PURPOSE:To easily execute operation by analyzing an object from inputted information, accessing a data base based upon the analyzed result, retrieving information necessary for the object, converting the retrieved result to a document format, and then outputting the document. CONSTITUTION:In an ordinary document preparing device, it is necessary for a user to form almost all the parts of a draft and input a large part of a document to be outputted. However, in this document preparing device, an analyzing means analyzes an object from information inputted from an input means and a retrieving means accesses a data base based upon the analyzed result of the analyzing means and retrieves information necessary for the object. An output means converts the retrieved information into a document format and outputs the document. Thereby, the information can be applied from the data base only by inputting minimum information required by natural language and a proper document can be prepared.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a document creation device that easily creates documents.

[Prior art]

Conventionally, when creating a document using a document creation device, you first create a draft, then enter the text one character at a time by operating the keys on the keyboard, or register common words and phrases, and input the document by calling up the common words and phrases. This created a complete document.

[Problem to be solved by the invention]

In conventional document creation devices, in order to create a document, the user must create most of the draft himself and input most of the document to be output, which requires a lot of time and effort. I had to give up.

[Means to solve the problem]

According to the present invention, an input means for inputting information, an analysis means for analyzing the purpose from the information input from the input means, and a database is accessed based on the analysis result by the analysis means to obtain the purpose. The above object is achieved by a document creation device having a search means for searching for necessary information, and an output means for converting the information searched by the search means into a document and outputting it.

〔Example〕

FIG. 1 is a system block diagram of one embodiment of the present invention.

In FIG. 1, IN is an input unit, which receives the user's natural language input through a device such as a keyboard, as shown in FIG. 7(a) INPUT, and transmits it to the CPU.

The CPU is a processing section, the details of which are shown in FIG. 2. By executing the DRAM, the functions shown in FIG. 2 are realized. Furthermore, on the memory, there is a knowledge base KB, a personal database IDE, and the data shown in FIGS. 3 and 6 (a).
, there is a program corresponding to the flowchart of FIG. 6(b).

In Figure 2, ANA is a natural language analyzer consisting of a parser (grammar analysis part), an understander (understanding part), and a plan contrastor (planning part), and it is a natural language analyzer that understands the general concept shown in Figure 5 (a). Knowledge base KBI.

Knowledge base KB2 of basic concepts in the fields shown in Figure 5(b)
, the user's personal database KB shown in Figure 5(d)
4. Using the general dictionary and grammar knowledge base KB5, act according to the process shown in Figure 4(a),
From the input information, the user's intention (for example, DI in FIGS. 7(a) and 7(b)) is inferred, and a plan (for example, J in FIG. 7(b)) is made. In this process,
The natural language analyzer ANA is managed by the management unit A described below.
interact with D.

AD can be used to search for specific information from the personal database IDE or to add new information to the personal database IDB.
This is the management section of the system that satisfies the requirements of the natural language analyzer ANA and natural language synthesizer SYs by writing in the .

The KB is an integrated knowledge base of the system, and the knowledge base KBI of general concepts as shown in Figure 5(a),
Knowledge base KB2 for basic concepts in the field as shown in Figure 5(b), Knowledge base KB3 for concepts of sentence synthesis as shown in Figure 5(c), Knowledge base KB3 for basic concepts in the field as shown in Figure 5(c), Knowledge base KB3 for language concepts with general dictionary and grammar knowledge. It includes a knowledge base KB5 and is accessed by the natural language analyzer ANA and natural language synthesizer SYS as required.

The IDB is the fifth database accessed and handled by the AD.
User's personal database KB as shown in figure (d)
It is 4.

In the KB and IDB having the configuration shown in FIG. 5, information related to a search term is derived in a lexicographical manner.

SYS is a natural language synthesizer consisting of a document preparer (sentence synthesis section) and a generator (sentence generation section), and receives the analysis results from the natural language analyzer ANA from the memory MEM, as shown in Fig. 4 (b). According to the process, knowledge base KBI of general concepts, knowledge base KB2 of basic concepts of the field, knowledge base K of concepts of text synthesis.
B3, operates using the knowledge base of language concepts KB5, K
Synthesize the sentence components according to the knowledge stored in B3, for example K in Figure 7(b), and output it as a natural language sentence, for example as shown in Figure 7(b) OUTPUT. and transmits it to the output section.

MEM is a memory that stores analysis results, lists, output documents, etc. by the natural language analyzer ANA.

Returning to FIG. 1, OUT is an output section that receives the output from the CPU and transmits it to the user through a display or an output device such as a printer or a speech synthesizer.

Next, the processing procedure of the system of the present invention will be explained with reference to FIG.

This system first uses natural language (NL) to generate parsers, understanzas, and plans.
Sent to a natural language analyzer that consists of a constructor. The processing process here is as shown in FIG. 4(a). That is, first, the parser uses the language concept knowledge base KB5, which has general dictionary and grammar knowledge, and the general concept knowledge base KBI, to first parse part by part, then clause by clause, and then concisely. The analysis is carried out step by step, and the input sentence is broken down into conceptual units, supplementing information if necessary, such as parts that are not explicitly stated or omitted, and then translated into a programming language. converted. This analysis becomes more accurate by considering previously input sentences. Next, in the understander, from among the conceptual units formed by the parser, the knowledge base KBI of general concepts, the knowledge base KB2 of basic concepts of the field, and the personal database K
B4 is used to search for an "actor" who is a person related to the action, such as the subject of the action or the object. Once the actor has been found, we then move on to understanding (searching for) the underlying goals and/or plans for the actor's relevant actions. Here, these goals and plans (or even the actors in some cases) may not necessarily be made clear in the input, and in that case, ``guess'' may be necessary during ``search.'' , select goals and plans for the actor from among the conceptual units obtained by the parser.

Next, in the plan constructor, among the goals selected in the understander, for those for which a plan to satisfy the goal could not be obtained under the understander,
Make a plan using KB2.

After this processing, if necessary, the user can output the created plan to the display of the output unit, printer, etc. Once output, the user can examine the output plan and evaluate the degree to which the system "understands" the user's intentions. Also, at this time, if necessary,
The user can modify and supplement the output plan, and proceed to the next stage using the revised plan as the plan for this stage. Here, the system writes the revised and supplemented plan together with the goal into the knowledge base KB2 of the basic concepts of the field, and learns this plan.

In the manner described above, the information obtained by the analyzer is sent to the synthesizer. The synthesizer consists of a text preparer and a generator, and its processing process is as shown in FIG. 4(b). That is, first, in the text preparer, the information obtained by the analyzer is organized using the knowledge base KB3 of the concept of text synthesis.

Also, here, the information necessary to adapt the above information to the concept of sentence synthesis stored in KB3 is
It is given from the knowledge base KBI of self and general concepts, the knowledge base KB2 of basic concepts of the field, and the personal database KB4.

Next, the "components" of the organized sentences are selected by the sentence generator in mode-on-speech (sentence types such as questions and requests, the degree of politeness, etc.), and are based on the KBI and language concept knowledge base KB5. The programming language is converted into a natural language by selecting words appropriate for the selected mode while following the grammar rules.

The natural language document created as described above is presented to the user by an output means such as a display or a printer. Here too, if necessary, the user can modify or supplement the output document, and the resulting information is
written to B.

Finish the work by printing the final result.

Hereinafter, with reference to FIGS. 6(a) and 6(b), the processing procedure performed by the natural language analyzer and text preparer when creating an English business letter using the document creation device of the present invention will be described in detail. In FIG. 6(a), upon receiving input from the user, the word "re" is entered in step 5100.
Search for "ply" to find out whether the letter you should write is a reply to a message from the other party.The word "r e p l"
If y''' is not found, it is regarded as a new transmission, and the process proceeds to step 5lot, where an actor is searched (estimated if necessary) from the input information and selected.Subsequently, step 51
In step 02, themes related to the actors selected in 5 lots are searched from the personal database IDB. Once the theme is searched, enter and IDB in step 5104.
Search for previous related information. If the information is obtained, the process advances to step 5105.

If the theme cannot be retrieved from the IDB in step 5102, the process proceeds to step 5103, where the theme is searched based on the input, and then the process proceeds to step 5105. In step 5105, information regarding the current situation is retrieved by searching or inferring from the input, and if information is obtained, the process proceeds to step 5106, where the user's goal is selected by searching or inferring from the input, and the goal is obtained. If so, step 5107 retrieves the plan from the input and IDB. Once the plan is obtained, a step 5108 searches for a goal lower than the goal, and if found, step 5107
Go back and search for plans for each goal.

Also, step 5104.5105.5106.510
In step 7, if no information is obtained through the search or inference, the process proceeds to step 5112 to request information from the user, obtain the information from the user, and then proceed to the next step.

Go back and if there are no subordinate goals in step 5IO8,
The process proceeds to step 5109, and it is determined whether or not this is the first call by considering it as the first call when there is no theme in the IDB in step 5102. If it is the first call, the process proceeds to step 5ilo, and steps 5103 and 5105
, 5106, 5107 is listed in the memory MEM, and if it is the first time, proceed to step 5ill, list the information obtained in steps 5104, 5105, 5106, 5107 in the memory MEM, and write it to the generator. send.

Next, if there is a "reply" in step 5100, the process advances to step 5201 in FIG. If the latest information ° is searched and the information is obtained, then proceed to step 5203.
The IDB is searched for the previous call, and if there is one, the purpose of the call, that is, the purpose of the sender's reply to the previous user's call is searched. Once the destination's calling purpose is obtained, in step 5204, a response to the destination's calling purpose obtained in step 5203 is searched from the input. Once this response is obtained, the process advances to step 5205. Also,
Step 5202.5203. In step 5204, if no information is obtained through the search, the process proceeds to step 5210 to request information from the user, obtain information from the user, and proceed to the next step, respectively. In step 5205, the purpose of the user's call other than the response is searched from the input, and if there is no purpose for making the call, the process proceeds to step 8206, and if there is, the process proceeds to step 5207. Step 520
7 and 5208, steps 5107 and 510
Similarly to Step 8, if there is a plan for the goal and a sub-goal, it and the plan therefor are obtained and the process proceeds to step 5209.

Returning to step 5206, step 5202.52
List the information obtained in step 04, and in step 5209 step 5202.5204.5205.5207.5
The information obtained in step 208 is listed. Step 5206
Or the information listed in 5209 is sent to the generator.

Next, with reference to FIGS. 7(a), (b), and FIG. 8,
Using a specific input example, the process of creating a transmission document from this input will be explained in detail. As an input example, Figure 7(a)
'write a 1etter to Sm1 as shown in
aboutvisiting Rocheste
r after the O8A meeting i
n April". Receiving this input information, the parser uses the knowledge of the KBI to find the unspecified verb (
When the subject of the clause (phrase) is the caller (5ender), and after the O3Ameeting is af
ter attending the O3A mee
ting and is compensated. Furthermore, the information is divided into conceptual units, * write to -Sm1thA, 5
ender -go to -Rochester-
AprilB, 5ender -go to -O3
A meeting-AprilC, (A) occ
urs after (B).

Next, in the understander, the actor and purpose are searched, and as actors, 5ender and Sm1th
was selected, and the purpose is A, which indicates the actions that should be taken in the future.
and B apply. Then, with information from KB4 (and the user if necessary), E, i, Sm1th i
s a professor of Univ, of
Rochester.

ii, sender-3mith relationship
ship.

5ender is working on a 1e
ns design system.

Sm1th's theory of aberrat
ion is used for 5ender'ss
system.

iii, O3A meeting is O3A a
annual meeting which will
held in US.

Get information about.

Next, from KH2, “Visit appropria
G
GOAL's PLA called et Information
Since it is included in N, F is a higher goal of objective A.
, 5ender -has -goal -of
-getting information by d
iscussing with Sm1th is obtained.

Next, search for a plan for the obtained goal.

For (F), first, (A) is the plan for (F),
Furthermore, as a plan for (A), from (÷) of E, G,
plan-for-goal (F):
go to Rochester (Univ, of
Rochester)=Ogo to US.

is obtained.

For B, from E(iii), H, plan -for goal (B): go
to US From above, (A) <' (F), 1 for (B), plan-for goals (A)
& (B): go to US is obtained.

Then, in the plan-funstructor, the plan for the actual action is obtained as follows.

J, plan: a, go to US.

b, do (B) (i, e, attend
OSA meeting)c, do (A),
(i, e, visit Rochester) then
In the document preparer, from the above, a, rela
tionship (5ender-Sm1th)
b, final goal (i, e, , (F
) )c, final plan (i, e, +
(J) ) is obtained as information. In addition, from the document synthesis knowledge base KB3, purpose of the 1
From the list of etter, visit person
Corresponding to d, 1nquire -plan executetab
le or not.

is added, and the same (from KH3, rhetoric such as the beginning and conclusion of a letter, i.e., e, greetings/gratitude, e
tc is added and each piece of information is arranged.

If you write down the results, Dear Prof, Sm1th,
(greeting) sender is w
orking on a single system
ender -Sm1th relation)Smi
th's theory of aberration
is used for 5ender's system
tem(sender-3mith relationship
) sender -has -goal of -di
scuzzing with Sm1th(final
goal) sender -go to -US
(final plan) attending OS
A meeting (final p
lan) visit Rochester (after
r the meeting) (final pla
n) sender-can-visit-or-n
ot? (inquire -plan execute
able or not) Looking forwa
rd to hear from you
(greeting) Yours 5incorely
(greeting).

Finally, the generator converts the above-mentioned document components obtained by the document preparer into a natural language document using the knowledge base KB5 of basic language concepts and taking into account the mode-on-speech mentioned above. converted.

As a result, a letter as shown at OUT PUT in FIG. 7(b) is output to an output unit such as a display or a printer.

〔effect〕

The present invention has the effect that by inputting the minimum necessary information (a small amount of information in natural language), information is added from the database and it is possible to create an appropriate document.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a detailed diagram of the CPU shown in FIG. 1, FIG. 3 is a diagram showing the processing procedure of the device configured as shown in FIG. 1, and FIG. Figure (a) is a diagram showing the processing procedure of the analyzer shown in Figure 2, Figure 4 (b) is a diagram showing the processing procedure of the synthesizer shown in Figure 2, and Figure 5 (a) is the content of the knowledge base KBI. Figure 5 showing
Figure (b) is a diagram showing the contents of knowledge base KB2, and Figure 5 (
C) is a diagram showing the contents of knowledge base KB3, Figure 5(d)
is a diagram showing the contents of knowledge base KB4, Figure 6(a),
(b) is a diagram showing the procedure for creating a letter, Figure 7 (a),
(b) is a diagram illustrating operations in response to specific inputs in the embodiment.

Claims (1)

[Scope of Claims] Input means for inputting information; analysis means for analyzing the purpose from the information input from the input means; and accessing a database based on the analysis result by the analysis means to achieve the purpose. A document creation device comprising: a search means for searching for necessary information; and an output means for converting the information searched by the search means into a document form and outputting it.