JPS5930169A - Document processing system - Google Patents

Abstract

PURPOSE:To edit a high-grade book with an index automatically, by allowing a document processing system to deform a keyword candidate into a keyword and generate the index together with given pages only by specifying the keyword candidate at the stage wherein a document is inputted to the document processing system. CONSTITUTION:An sentence or phrase to be employed as a keyword in the document is specified by an operator during input operation together with added control codes (the starting and ending of the document, starting and ending of reading, coined word, and word to be modified). A word dictionary 18 is used to make a grammatical analysis of an extracted sentence (keyword) specified by the control codes. The dictionary contains keywords arranged in reverse reading order, a concatenation table of annexed words, derivative nouns of terms, and an inflection table. The extracted sentence is divided into clauses by the grammatical analysis and then the deformation into the keyword is performed by two kinds of deformation pont, i.e. deformation into a derivative noun and deformation by the specification of a word to be modified.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field of the Invention The present invention relates to a document processing system, and is intended to provide the system with a function of generating a table of contents and an index entry from a specified sentence in a sentence.

Conventional technology and problems In the Hisai processing system, if Hisai is simply input continuously (in a hexagonal pattern) along with the control code on the magnetic medium, W
IWIJ31. is typesetting processing and page breaks every page with a predetermined number of lines, line breaks are made after a predetermined 93 minutes, a part of the text that has been set for 1 hour is extracted along with non-pulls from that page to create a table of contents and an index.
etc. can be edited.

However, in the conventional Hisai processing system, the headwords in the table of contents and index are the exact words in the main text, and editors cannot make changes such as inflection or inversion of word order, as they would do manually. For example, headwords are converted into nouns, such as ``input by card,'' but the words in the main text are often sentences themselves, such as ``input by card,'' and are used as headwords. This results in unnatural results. For the sub-game ζ, choose a word or phrase from the main text that is suitable for the headword, or if possible, do it manually (1y correction is necessary, which is troublesome).

OBJECTS OF THE INVENTION Therefore, the present invention aims to transform a headword candidate sentence specified in the main text into a suitable word or phrase as a headword (1) to add function to the Hisai processing system.

Structure of the Invention The present invention is a Hisai processing system that performs pagination LJ on the main text of a manually-written document, and also creates and edits a table of contents, index, etc., in which words are arranged in reverse reading order, and examples and subordinate words are The conjunctive table is used, the conjugated noun is used for Yo-kun, the word Jikichi is used for the conjugation table, and the control code indicates that it is a headword candidate, that it is a coined word, and that it is a modified word. In response to the above-mentioned text, which is instructed to be a ``descriptor'', the entry word candidate is converted into a transposed noun, and if there is an ``e'' modification, the relevant clause is modified into an ``e'' and the remaining clauses are modified into an adnominal. It is characterized by comprising a processing device that generates a headword by transforming it into a word and creates an index together with the searched pages. do.

Embodiment of the Invention Fig. 1 shows the flow of processing in the LJ processing system, where 10 is a document, 12 is a keyboard of a data encoding device, etc., 14 is a floppy disk, and 1G is an AL'W machine. This is a processing device. To process the sentence P4, the operator operates the keyboard 112 to input the contents of the document 10 into the hard disk 14 in the form of a T-15 format, while inputting the required control code. The processing device 16 outputs the floppy disks 14 one after another, and uses the control code (jll means page assignment, line break, "I-order creation, etc., to create a copy. Index creation involves page assignment, etc.") and instructs the text to be performed for this text. The terms listed in the index and the pages in which they exist are extracted, arranged in alphabetical order, etc., and page assignments are made to create a copy of the index page. It attempts to provide first-class grammar analysis and headword transformation functions.

The diagonal a1 frame portion in FIG. 1 shows the portion according to the present invention during index processing. In order to transform the extracted sentence into a headword, it is necessary to change the control fffll: J M to a famous person. Therefore, the present invention will be explained below in the order of input method, analysis of extracted sentences, and transformation of extracted sentences.

Human power method: sentence t! The sentence or phrase to be adopted as the headword in F is specified by the operator when manually inputting the original document to a floppy disk. The specification procedure is as follows. Please \deO
is the beginning of a sentence, (◎ is an open line, O is the beginning of a reading,
O is a control code for opening.

Qt+l+: l-end O Kunmatsu (I read the data a'j
Input 0 Yomicom OO In this example, the sentence in the main text is [read data from the terminal], and when the operator inputs the main text into the floppy disk, the operator controls the above text marked with (=J) by adding an underline, etc. Codes 0, 0, etc. are entered every month.As shown in this example, for kanji, enter the reading along with the kanji.The reading is entered immediately after each word for kanji that is only kanji or contains kana. A combination of multiple words, such as "terminal device", can be used as one word.rlil terminal device Otan "7TSOUJI・" or separately as terminal O Kunmatsu O device 0 = Hashichi O. Good. Kanji words that are mixed with makokana can be written separately for each kanji. For example, "to read a card" can be written as "to control a card 0 yo ◎ take ◎ 1 ・Oru". Control code 1' The kanji that precedes Q (the ones up to the beginning of the sentence or the O immediately before it) are
It is treated as having a reading starting from .

Since the index mainly lists Hisai's unique usages, it often deals with coined words (mainly nouns) that are not included in the word dictionary. For such coined words, control when inputting:,
It is clearly stated in the text according to the month of J-1'. for example"
J'l! for the coined word "zabmitsuto" in the sentence "zabmitsuto". [At the beginning of RA, open each control code 0.0 and perform a 1-0 job. ” said J.

Headwords are often made into names, but there are 17 + 1 nouns, pronouns, or numerals such as [-to take the card, to display on the screen, to be operated by the operator, etc.] , if it is a simple sentence with one verb, adjective, or adjective, the predicate in the idiom can be used as a transposition noun (if it exists)
, and the remaining segments can be used as modifiers related to transposed nouns, such as ``reading a card'', ``displaying on a screen'', or ``operating an operation'', which is the basic form of heading doubling. For transformations that are suitable for this basic form, we will not give instructions on how to transform them, specifically, we will not explain the control code (unnecessary).
. However, if the part other than the predicate of the sentence has multiple circularity, such as a subject or an object, all of them are 1-to-modifiers and i
! 1rT, so decide which one you want.

The predicate is transformed into an adnominal modifier that depends on this predicate.
Q, and ζ are determined by 1 digit. For example, in the sentence 1-The operator cancels the job, the predicate r =
f (1st) other than "l-yansel" has the nominal word "meberator" and "job", both of which are subject (lk hQi
i hatari 1! I specify it as ``RQ Operator O cancels the job'' or 1-Operator G,
・Cancel job 0]. In this way, the processing device transforms the former into "the operator who cancels the job" and transforms l&fj into "the job which is canceled by the operator". There may be other ways to transform the entry into a headword, but in the present invention, the above-mentioned transformation into a noun,
Assume two methods for determining one modified word (false).

Grammar analysis of extracted sentences: The authors with indexes are specialists, practical stores,
It can be said that it is quite specific in articles and the like, and there are no long sentences among the words and phrases used as eternal farewell. To enumerate its general characteristics, it is a declarative sentence written in ordinary ``1'' words, and structurally it is a simple sentence. ■The relationships between clauses are only those between subjects and predicates, and between (1 and ornaments and modified objects. 1 adds a special meaning to the 0 sentence (=J 111. Quantity 1, auxiliary verbs such as katsushi, hearsay, etc. are not used.■ Adjunctive adverbs are not used. In the present invention, -no sentences with such characteristics are also selected. Assume that

i1i'l fff++ code 0. Q (the former stands for Sentence Star 1-1f, LW stands for Sentence End)
To grammatically analyze an extracted sentence defined as 1 in 1'! Use Glossary 1B. The word dictionary 18 is a general Japanese dictionary (I
The selected words are arranged in alphabetical order by taking into account the characteristics or limitations mentioned above and the specialized field of the document, and each word is marked with its crystal E, type of conjugation, etc. The difference from general Japanese dictionaries is that all entry words are arranged in reverse reading order.

For example, ``Yometoru'' (read) is expressed as (-to no yo). ?I′f is 1-
The fourth from the end is registered as the position that should be occupied by the word ``yo.'' The reason for arranging them in reverse reading order is that the retrieval ref rate improves because of the IIN characteristics of Japanese, such as the fact that predicates come last.

Also, in the word Jikichi, 1! A connection table is prepared for I1 idioms, ie, suke riiiil and suke ni1J. For example, auxiliary verb 1
The connection table for This table shows the auxiliary verb ``areru'' (=Jkuiha is the verb Uj narazu line declension conjugation 2, upper 1st conjugation, lower 1st conjugation, and power declension conjugation each unnatural form, auxiliary verb 1-1!ru). difference·
If it is ``keru'', then the unformed form shows that J) ru. Sentences used as headwords have the above-mentioned z1η characteristics or restrictions, so there are not many edicts ("J'W'r").With this connection table, when the auxiliary verb "areru" is detected, its It is easy to conclude that the antecedent word is consistent with the unformed form in the table above.

There are multiple forms of kanji okurikana, such as reading, reading, hitting, hitting, etc., so a specific format is not specified, but it is sufficient to match the overall reading and whether the kanji is mixed with kana (search can).

Well, every month I put a transposed noun in each noun. for example"
``Read'' is added to ``to control.'' If there is a negative form of the noun, such as 1-minute 14+1 J ni kariru 1-non-separation, add 1 to this as well.

Also, "when executing", "when outputting", "when inputting", "when leaving in the afternoon", when, when, in the afternoon, etc. (=J conjunction modifier clause without accompanying Create name Rn
Please indicate this to J. This means that the conjunctive modifier converts the predicate into 1m f! t) IL, when doubling the heading of a sentence, the adjunctive modifier is transformed into an adnominal modifier, which is useful at that time. For example, if you write ``1-I'm leaving in the afternoon'' and ``I'm leaving in the afternoon'', it becomes a headword, but this is because the inflection [-departure] is used as ``departure'', and the conjunctive modifier that modifies the predicate is ``departure in the afternoon''.
``Afternoon'' is changed to the adnominal modifier ``afternoon'' and is i! 7. Also, these meibancho sometimes take the form ``1-pm'' with ``suke1iiJ l''. In this case, the adjunct t+￥ is also converted into a ``+'1-f&'', but when the direction is J-, it can also become J to 1゛～, The above-mentioned 1-Indication to that effect provides a finger for identification. Furthermore, converting a conjunctive modifier to a conjunctive modifier simply means adding 薖えPO [no].
In the second example, 1 - "When executing", "When outputting",
``When using human power.''

Next, there are various conjugations of predicates, etc. in the extracted sentences; on the other hand,
A final form is desired for headwords, but this conversion is done using a conjugation table. The conjugations of pragmatics and auxiliary verbs are shown below.

As shown in this table, the conjugated form is the proximal conjugation for the declarative,
I have them for different types, such as homophones and joichidan conjugations, and since auxiliary verbs are irregular, I have them for each word. This table is sufficient for headwords.

Since each character in the extracted sentence part and the analyzed sentence is treated sequentially from the end of the sentence, they are stored in an area arranged in the order of character phase as shown in FIG. Extracted sentences include perfect final sentences that end with the final form of a pragmatic/auxiliary verb, and incomplete final sentences that end with a conjunctive particle or a conjunctive cessation of a pragmatic/auxiliary verb (extract headwords from these two types) ) is preferable for the grammatical analysis of the extracted sentence and a J-complete sentence, so this conversion is performed in the sentence-end processing. The conjunctive particles are ba, demo (but), ni, so, te (de), nakara1. It may be said that it is a matter of degree (dari) or something, and these processes are performed as follows. This character (11 refers to the last character of the sentence shown in Figure 2).

(1) When character (1) - “ba”, “no”, “mo゛, or ゛rapa”, it becomes a conjunctive form of a pragmatic/auxiliary verb and does not have a stomach, so the conjunctive particle “ba”, “ Mono no pa, ゛mo (but)'', and ゛nagara'' are the endings of the words.

(2) Character (1) - “′ni”, “te(de)” or “su”
In this case, there is a possibility that it is the conjugated ending of the word, so search the word dictionary for the final ending from the conjugation table (the search method will be described later). In a dictionary, each conjunctive particle, “noni,”
te (de) °°, “tari (tari)”.

(3) In the case of conjunction dij, check the connection pattern from the connection table for each word, find the final ending from the conjugation table,
The words to be connected are determined by the word dictionary.

(4) In the case of pragmatics and auxiliary verbs, find the final ending from the conjugation table and determine the word using the word dictionary.

In this way, all extracted sentences are complete stop sentences, and grammatical IW analysis is performed on them. First, assume that the character ■+11 is a word and look it up in a dictionary. Dictionaries are read in reverse order, so if you look at it, you'll see the corresponding one ((J), then the letter (2).
select the same one, and also select the same letter (3),
If you repeat the operation ζ;<, you will come across some words that can be considered words. These are trees representing word connections (1
−Lee) is the root of the structure.

Figure 3 shows this situation. Word w,,,w+,,...
... is a group of words connected to word W1, word w, , ,
lol...

W122... is the word W++, W+2・
The fli word JIY is connected to... In addition, the character ``
Immediately before ``wo'', the boundaries between katakana, hiragana, alphanumeric characters, commas, and the start and end points of coined words, etc., are word breaks, so these serve as clues when cutting out words. ■As a result of looking up a dictionary, if you find that it is a (tJ genus of an auxiliary NHJ, auxiliary verb, etc.), check the connection table mentioned above, and draw up a conjugation table based on the final form classification and conjugation of the connecting words, and Return to shape.

Note that this is not necessarily determined at will. Next, based on the remaining character string extracted as a li word in the previous step or the character string whose ending has been returned to the final form in the previous step, a ``word'' is drawn as in the above step . This determines the next word. The same process is repeated thereafter. ■When performing step (2) above, the process may get stuck due to a word that does not appear in the dictionary, even though it is not a coined word. In that case, that 'branch' is considered to be dead and processing is stopped, and processing moves on to the next branch. ■ After processing all branches, if there are multiple living branches, human In simple cases, the word order is determined according to the meaning.Since the extracted sentence is a kanji sentence mixed with kana, the word sequence is determined without being confused by homonyms etc. The conversion rate is high.Once a word is determined, its details can be found in a dictionary, so for each word, extract it when looking up the dictionary and write the part of speech for each word as shown in Figure 4. (In the case of conjugated words, the conjugated form is also included.) Pl, P2...... are part of speech 12, part of speech 2, etc.
Does not indicate... ■From the sequence of parts of speech, “independent word → - dependent word sequence” is a phrase, so recognize this phrase. Each clause is marked 1a by corresponding word numbers as shown in FIG. In this way, the extracted sentence is divided into clauses.

Transformation of extracted sentence: After dividing the extracted sentence into clauses as described above, the next step is to recognize the function of each clause. As mentioned above, headwords are quite specific as sentences, so the functions of clauses are small, such as subject and predicate, conjunction or adnominal modifier, and modified word. Since the extracted sentence is in the perfect final form, the predicate can be a clause at the end of the sentence. However, the following ■,
In the case of ■, there is a word that expresses the meaning as a predicate in the preceding clause, so these are grouped together as a continuous clause. ■When the clause at the end of the sentence is a part-of-speech classification auxiliary predicate, such as ``Iru'' in ``1-I'm reading a card,'' ■[File oniving j's ``Mapun J, ji゛variation i1i' J
A clause consisting of the word ``meito'' that precedes a phrase containing ``r.''. The former uses ``yomi-no-tori-tu'' and the latter uses ``open'' as a conjunctive clause predicate. Next, as for the subject, since the extracted sentence is a simple sentence, that is, a sentence with one subject, the subject is a clause accompanied by case particle RHJ ``ga'' or the adverbial particle [-wa].

Since conjunctive modifiers generally take the following form, these clauses are considered conjunctive modifiers. ■Noun → −Hiragi particle (“ni”
, U to”, ′to”, “and “gara”.

゛Yori”, “De”, etc.), ■verb (conjunction) 10, Q)
Yomei/Support WiiJ (Conjunction) Ura Conjunction of Ten Conjunction Particle (
+ and decorations (゛te(de)'', ゛nagara, ``tari゛, etc.), ((i) adjective do] on the conjunctive side 1liiJ method (adjective 2 of the final classification), (CJ) adjective verbs Conjunctive adverbial method (verb 3 of the final form classification), (D example) Conjunctive without a genitive (l and a noun that forms an ornamental clause (when, when, ゛toki).

``ori'', ``morning'', °'afternoon'', etc.).Adnominals (1 and decorations are based on the above-mentioned constraints of extracted sentences, i.e., the relationship between clauses is due to the relationship between subject-predicate or modification, and the relationship between arguable and decoration). Because there is, t);1
Items (clauses) other than subjects, predicates, and adjunctive modifiers are adjunctive modifiers.

In the case of transformation by transposition name NHJ, the predicate part is transformed into a noun, so the adjunctive modifier related to the predicate is transformed into an adnominal modifier 8h. Although it is necessary to recognize the modification relationship tjfg, it is difficult for the processing device to recognize the meaning based on the meaning, so this is done based only on the positional relationship. That is, the continuous flh fifir word is 1
It is assumed that the closest repeated term is used. for example"
When executing a job, it is assumed that in the sentence 4'S (177) 1, ``job'' modifies ``execute'', and ``when'' is [``1111 prepare J''. Similarly, “
In ``prepare before execution'', 1゛in advance of execution'' modifies ``prepare'', and ``prior to'' modifies ``prepare'', and 1-
"Prepare a job to be processed in the afternoon" means "1-afternoon" is 1-
"to process" is modified.

Once the clauses and their modification relationships are known, they can be transformed into headwords. The transformation method is transformation by transposition nounization and #E
There are two types of transformations based on the fili word specification.

Transformation by transmutation name vyJ is performed as follows. nu et al.
■Find the transposed noun of the predicate in the predicate clause.

As mentioned above, Jiyoi lists the transposed nouns for each noun.

If this clause contains a negating auxiliary verb such as ``nai'', the transposition name diJ will take the negative form.

■The particle “ga” or ゛wa゛ in the subject clause of the case particle °′
Replace with ■ Modify adjunctive modifiers that follow predicates
te is on the verge of recovery. Dirt is quite troublesome and cannot be done uniformly, so do it ff/I separately as follows.

a. Change the auxiliary ayJ “to” into the particle “”. particle pahe”
, °" and °°, ゛kara(yori), ゛de", °te, etc., add ``no'' after it.

Example: Data driving by hand - Personal data Driving on a road - Ahead of a road driving data error - Ahead of a data error - Ahead of a data error - Ahead of a data error - Checking while displaying data on a data controller -・Check while displaying the data], the particle ° “ni” is converted as follows.

・1 (11) Create a continuation fife + word clause without yakugo Replace ``to the pa attached to the noun'' with ゛ of ゛.For example, ``When executing a job, make a partial change'' means 1゛execute the job. Change the case.

・When there is an auxiliary verb (reru, wareru, seru, let) in the predicate, the ``ni'' in the conjunctive modifier is replaced with ``no''. For example, ``I--make the operator make a trench.'' is set to 1-operator's side work', and [-specified by user] is set to 'user's designation'.

・Other than the above, "ni" often indicates a direction or place, so replace it with "to".For example, [-gather at the playground] is changed to "exercise" -gathering to -1, and [display on one screen] ”
1.Display on screen" and "-Request to operator"
is 1 - 1 request for operation.

C5 Adjunctive forms of adjectives and adjective verbs are converted into adnominal forms using the conjugation table.

d. Add “no” (=J to nouns (nouns that form a conjunctive modifier clause without the aforementioned <oi=J genus). For example, [
``Start input processing in the afternoon'' means ``start of manual processing of l'll&''.

Transformation by specifying a modifier is relatively simple; all you have to do is bring the specified modifier at the end and make the rest its (1-t modifier). Specifically, the conjugated ending of the predicate is It is converted into an adnominal form according to the conjugation table, and the specified modified word follows it.The example genitive that was connected to the modified h (11 words) is discarded.As an exception, the adnominal form following the noun after the auxiliary verb is Something without body shape (lIJi definite auxiliary verb ゛°゛) is specially changed (changed to “is”).Also, the auxiliary aHI that represents the subject
If there is a ``ha'' in ``ha'', change it to ``ga゛''.

Example: User manually enters an operand from the console.

(same decoration word Variant word user → User console where the operand is input from the console − Console where the user inputs the operand −
Luobelan 1--Operan that the user inputs from the console.In this way, a specified word or phrase in a document can be extracted, converted into a form suitable for a headword, and an index can be created.

[9] Results of the Invention As explained above, according to the present invention, a headword candidate is specified at the stage of inputting a word to the sentence processing system, and the Daikichi processing system then converts the headword candidate into a header. Since the index is created along with the page that is searched, operator intervention is reduced, and the text processing system can automatically edit the advanced figures on the index side. .

[Brief explanation of drawings]

FIG. 1 is a flowchart 1 showing the process flow of the Daikichi processing system to which the present invention is applied, and FIGS. 2 to 5 are explanatory diagrams of the processing procedure. In the drawing, 0 is a document, 12 is a display keyboard, 14 is a floppy disk, 16 is a processing device, and 18 is a word dictionary.

Claims (1)

[Claims] In the Hisai processing system, which paginates the main text of the input text, creates a table of contents, index, etc., and performs °ζ editing, the words are arranged in reverse reading order, Conjunction tables are used for words, transposed nouns are used for pragmatics, and words with /iIi table are used for pragmatics and auxiliary verbs.The control code indicates that the word is a headword candidate, that it is a coined word, and In response to the above text that was instructed to be a modified word, some headword candidates were converted to transposed nouns, and for those with a modified word specified, the relevant clause was used as a modified word, and the remaining clauses were changed to an adnominal modifier. 1. A document processing system comprising: a processing device that transforms into °C, generates a headword, and creates an index together with the cut out page.