JPH0330183B2 - - Google Patents
Info
- Publication number
- JPH0330183B2 JPH0330183B2 JP58182229A JP18222983A JPH0330183B2 JP H0330183 B2 JPH0330183 B2 JP H0330183B2 JP 58182229 A JP58182229 A JP 58182229A JP 18222983 A JP18222983 A JP 18222983A JP H0330183 B2 JPH0330183 B2 JP H0330183B2
- Authority
- JP
- Japan
- Prior art keywords
- dictionary
- sentence
- word
- sentence generation
- rule
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
- 238000010276 construction Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 9
- 239000002245 particle Substances 0.000 description 4
- 238000000034 method Methods 0.000 description 2
- 241001672694 Citrus reticulata Species 0.000 description 1
- 235000016496 Panda oleosa Nutrition 0.000 description 1
- 240000000220 Panda oleosa Species 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
- G06F40/55—Rule-based translation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Machine Translation (AREA)
Description
【発明の詳細な説明】
〔発明の技術分野〕
本発明は、意味ネツトワークを入力情報とする
自然言語文章生成装置に係り、特に構文規則辞書
とともに、特定単語間に存在する特別な用法を登
録した共起関係辞書、隣接関係辞書を用い、単語
の内容に依存する慣用句、語呂等の自然言語の特
性情報を全て辞書化することにより、文章生成処
理の本体機構を任意の言語に共通化できるように
した自然言語文章の自動生成方式に関する。[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a natural language sentence generation device that uses a semantic network as input information, and in particular, registers special usages that exist between specific words together with a syntax rule dictionary. By using the co-occurrence relationship dictionary and adjacency relationship dictionary, we can make the main mechanism of sentence generation processing common to any language by converting all characteristic information of natural language such as idioms and rhymes that depend on the content of words into dictionaries. This paper relates to a method for automatically generating natural language sentences.
たとえば英語を日本語に翻訳する機械翻訳装置
においては、英語文を計算機に認識できる意味ネ
ツトワークと呼ばれる形式に変換した後、文章生
成装置に入力して日本語文を出力するようにして
いる。
For example, a machine translation device that translates English into Japanese converts an English sentence into a format called a semantic network that can be recognized by a computer, and then inputs it to a sentence generation device to output a Japanese sentence.
第１図は、意味ネツトワークの１例を示す。意
味ネツトワークは単語「私」、「会う」、「少女」、
「美しい」からなるノードと、特定の２つの単語
間の関係を表わす記述＜Statement＞、＜actor
＞、＜Object＞、＜past＞からなるアークによつて
構成されている。アークの例としては他にも＜
time＞、＜place＞、＜goal＞、＜比較＞などがあ
り、このような意味ネツトワークの構造は、任意
の言語に対して不変であるという性質をもつてい
る。 FIG. 1 shows an example of a semantic network. The semantic network consists of the words ``me'', ``meet'', ``girl'',
A node consisting of “beautiful” and a statement expressing the relationship between two specific words, <actor>
It is composed of arcs consisting of >, <Object>, and <past>. Other examples of arcs include
time>, <place>, <goal>, <comparison>, etc., and the structure of such a semantic network has the property of being invariant for any language.
ところで、従来の文章生成装置は、プログラム
によつて意味ネツトワークから自然言語文を生成
していたが、その場合、英語用、日本語用などの
ように、生成しようとする各言語ごとに異なる生
成装置が必要とされた。 By the way, conventional sentence generation devices generate natural language sentences from semantic networks using programs, but in this case, the sentences differ depending on the language to be generated, such as English or Japanese. A generator was required.
しかも従来では、たとえば日本語文章生成の場
合、単語間に挿入すべき助詞や、時刻、命令形や
義務形などのスタイルを変形させる場合の変化語
は、人間が判断してキーボードから入力して教え
る必要があり、その他の言語の場合も、計算機の
みを利用すると、不自然な文章が生成されること
が多く、全く自動的に文章を生成することはでき
なかつた。これは、従来の装置では、辞書が構文
中心方式で作成され、意味、慣用語、語呂などに
もとづく、たとえば「山へ上る」と「山へ登る」、
「階段を上る」と「階段を登る」などについての
正誤の区別や、「私がリンゴを食る」、「彼はミカ
ンを食べる」、「リンゴは私が食る」などの助詞の
適切な選択ができなかつた。 Moreover, in the past, when generating Japanese sentences, for example, particles to be inserted between words, time, and inflectional words to change the style such as imperative or obligatory forms were judged by humans and inputted from the keyboard. Even in the case of other languages, using only computers often produces unnatural sentences, and it has not been possible to generate sentences completely automatically. This is because in conventional devices, dictionaries are created in a syntax-centered manner, and are based on meanings, idioms, and puns, such as ``to climb the mountain'' and ``to climb the mountain.''
Distinguish between correct and incorrect statements such as "climb the stairs" and "climb the stairs," and the appropriateness of particles such as "I eat the apple,""he eats the mandarin orange," and "I eat the apple." I had no choice.
本発明の目的は、単語間の共起関係、構文パタ
ーンにしたがつて、質のよい自然言語文を自動的
に生成することにあり、特に言語に依存しない生
成方式を提供することにある。
An object of the present invention is to automatically generate high-quality natural language sentences according to co-occurrence relationships between words and syntactic patterns, and in particular to provide a language-independent generation method.
本発明の要点は、文章生成装置において、生成
のメカニズムの部分と規則の部分とを分離するこ
とにより、多種の言語の文を生成する際にメカニ
ズムは共通で、規則の部分だけを与えなおすだけ
ですむようにし、規則は構文を制御する規則とし
て構文規則辞書、単語の選択、言い回しを制御す
る共起関係辞書、形態素生成を制御する隣接関係
辞書を用意し、従来の構文中心方式に対して単語
の意味や慣用的用法による文の変化を考慮できる
ようにすることによつて、質のよい自然言語文を
生成するようにしたものである。
The key point of the present invention is that by separating the generation mechanism part and the rule part in a sentence generation device, when generating sentences in various languages, the mechanism is common and only the rule part is re-supplied. We have prepared a syntax rule dictionary to control syntax, a co-occurrence relationship dictionary to control word selection and phrasing, and an adjacency relationship dictionary to control morpheme generation. By making it possible to take into account the meaning of sentences and changes in sentences due to idiomatic usage, it is possible to generate high-quality natural language sentences.
そして本発明の構成は、それにより、言語に依
存しない普遍的な中間表現である意味ネツトワー
ク形式の単語情報入力手段と、文生成規則情報フ
アイルと、文生成処理装置とをそなえ、上記文生
成規則情報フアイルは、生成目標言語側の単語辞
書、横文規則辞書、共起関係辞書および隣接関係
辞書を含み、上記文生成処理装置は入力された意
味ネツトワーク形式の単語情報に含まれる各単語
について単語辞書および構成規則辞書を参照して
文章の基本的な構成範囲を設定するとともに、共
起関係辞書および隣接関係辞書にもとづいて２つ
の単語間における最適な表現を選択し文章を決定
することを特徴とするものである。 Accordingly, the configuration of the present invention includes a word information input means in the form of a semantic network, which is a universal intermediate expression independent of languages, a sentence generation rule information file, and a sentence generation processing device, The rule information file includes a word dictionary for the target language, a lateral sentence rule dictionary, a co-occurrence relationship dictionary, and an adjacency relationship dictionary, and the sentence generation processing device analyzes each word included in the input word information in the semantic network format. In addition to setting the basic composition range of a sentence by referring to a word dictionary and a composition rule dictionary, the sentence is determined by selecting the optimal expression between two words based on a co-occurrence relationship dictionary and an adjacency relationship dictionary. It is characterized by:
以下に、本発明の詳細を実施例にしたがつて説
明する。
The details of the present invention will be explained below with reference to Examples.
第２図は、本発明の１実施例装置の構成図であ
る。図中、１は意味ネツトワークデータ、２は文
生成処理部、３は生成言語選択部、４は文生成規
則情報フアイル、５は日本語文生成規則、６は英
語文生成規則、７は露語文生成規則、８は単語辞
書、９は隣接関係辞書、１０は構文規則辞書、１
１は共起関係辞書、１２は推論部、１３は推論規
則辞書、１４は自然言語文を示す。 FIG. 2 is a configuration diagram of an apparatus according to an embodiment of the present invention. In the figure, 1 is semantic network data, 2 is a sentence generation processing unit, 3 is a generation language selection unit, 4 is a sentence generation rule information file, 5 is a Japanese sentence generation rule, 6 is an English sentence generation rule, and 7 is a Russian sentence Production rules, 8 is a word dictionary, 9 is an adjacency relationship dictionary, 10 is a syntax rule dictionary, 1
1 is a co-occurrence relationship dictionary, 12 is an inference section, 13 is an inference rule dictionary, and 14 is a natural language sentence.
本実施例装置は、自然言語文生成を行うための
規則の集合である辞書群８，９，１０，１１と、
それらの規則によつて自然言語文を生成する文生
成処理部２を主要な構成要素としている。 The device of this embodiment includes dictionary groups 8, 9, 10, and 11, which are a set of rules for generating natural language sentences;
The main component is a sentence generation processing section 2 that generates natural language sentences according to these rules.
文生成処理部２は、生成したい部分の意味構造
を表現する意味ネツトワークデータ１を入力とし
て受け取り、上記の各規則を参照しながら、対応
する自然言語文を生成する。生成する自然言語の
種別は、生成言語選択部３に生成言語指定信号を
送り、規則５、６、７等を選択することにより決
まる。また、文を生成する際、生成処理部のもつ
論理機能では処理できない特殊ケースについては
必要に応じて推論部１２および推論規則辞書１３
によりバツクアツプする。 The sentence generation processing section 2 receives as input the semantic network data 1 expressing the semantic structure of the part to be generated, and generates a corresponding natural language sentence while referring to each of the above rules. The type of natural language to be generated is determined by sending a generation language designation signal to the generation language selection section 3 and selecting rules 5, 6, 7, etc. In addition, when generating a sentence, the inference unit 12 and the inference rule dictionary 13 are used as needed for special cases that cannot be handled by the logical functions of the generation processing unit.
back up.
生成する言語は、生成規則として日本語規則と
接続すれば日本語、英語規則と接続すれば英語…
…のように生成規則を取りかえるだけで目的の言
語が生成できる。 The language to be generated is Japanese if you connect it with a Japanese rule as a production rule, and English if you connect it with an English rule...
You can generate the desired language by simply changing the production rules like...
規則を構成する単語辞書８、隣接関係辞書９、
構文規則辞書１０、共起関係辞書１１は、日本語
文生成規則５内にのみ示されているが、各言語ご
とに作成されている。 A word dictionary 8 constituting rules, an adjacency relationship dictionary 9,
Although the syntax rule dictionary 10 and the co-occurrence relationship dictionary 11 are shown only in the Japanese sentence generation rules 5, they are created for each language.
単語辞書８は、各単語の見出し、表記、意味記
号、構文記号、隣接情報、特徴から成る。 The word dictionary 8 consists of headings, notations, semantic symbols, syntactic symbols, adjacent information, and features of each word.
第３図は、英語単語辞書の例を、単語「give」
と「by」について示す“表記”は、英語では一
般に“見出し”と同じであるが、日本語の場合、
“見出し”はカナ文字で、“表記”は漢字となるこ
とがある。まだ“隣接情報”は３人称単数現在の
語尾が、Ｓか−esかなどを表わす。 Figure 3 shows an example of an English word dictionary for the word "give".
The “notations” for “by” and “by” are generally the same as “headings” in English, but in Japanese,
The “heading” may be in kana characters, and the “notation” may be in kanji. The "adjacent information" also indicates whether the current ending of the third person singular word is S or -es.
隣接関係辞書９は接続マトリツクスで表現さ
れ、単語と単語が隣接し得るかどうかを示す。例
えば英語においては、
Γ不定冠詞はａとanのどちらをとるか、
Γ３単現で接尾辞は−Ｓ、−es、……となるか
Γ進行形で接尾辞は−ing、ling、−ping、……と
なるか等の形態素生成の規則も表現する。 The adjacency relationship dictionary 9 is expressed as a connection matrix and indicates whether words can be adjacent to each other. For example, in English, does the Γ indefinite article take a or an? It is Γ3 singular and the suffixes are -S, -es, ... or it is Γ progressive and the suffixes are -ing, ling, -ping. , etc. It also expresses rules for morpheme generation such as .
構文規則辞書１０は、意味ネツトワークから自
然言語文に交換するための規則である。第４図に
その具体例を示す。 The syntax rule dictionary 10 contains rules for exchanging natural language sentences from a semantic network. A specific example is shown in FIG.
共起関係辞書１１は、前置詞の選択や、言い回
しの制御を行う。第５図にその具体例を示す。図
の例は、make houseとは言わず、build
houseとは言うことを示し、図の例はrobの対
象物には前置詞ofがつくことを示している。 The co-occurrence relationship dictionary 11 selects prepositions and controls phrasing. A specific example is shown in FIG. The example in the diagram does not say make house, but build
The example in the figure shows that the preposition of is attached to the object of rob.
文生成処理部２は、与えられた意味ネツトワー
ク中で＜Statement＞のアークが入つているノー
ドから処理を始める。第６図に示す例の場合は、
「give」がそれである。以後、かつこ内は第６図
の例における場合を示す。（なお、第６図の例は
“彼は本を買つて彼女にあげた”を意味表現して
いる。）まず、そのノードの意味記号から対応す
る単語を単語辞書８よりひいてくる（give、
present等の単語がひける）。次にその候補単語か
ら１単語を選び、その単語の構文規則を適用す
る。giveが選ばれたとすると、構文規則DVが
giveのノードに対して適用される。第４図の構文
規則によつて、まずvheadの規則が適用される。
実際にはここでgiveの過去形のgaveの選択、文
頭副詞や複文の従属節の処理がなされる。次に
act headの構文規則でheが、vgで自分自身gave
が生成され、＜goal＞でsheの目的語her＜object
＞でbook which he bought to herが生成され
る。この生成の際共起関係辞書によつて共起関係
（gaveに対して＜goal＞は前置詞をとらず、目的
語のherが接続するなど）が検査される。 The sentence generation processing unit 2 starts processing from the node containing the <Statement> arc in the given semantic network. In the example shown in Figure 6,
That is "give." Hereinafter, the case in the example shown in FIG. 6 will be described. (The example in Figure 6 expresses the meaning of "He bought a book and gave it to her.") First, the word corresponding to the semantic symbol of that node is retrieved from the word dictionary 8 (give ,
Words such as present can be found). Next, select one word from the candidate words and apply the syntax rules for that word. If give is chosen, then the syntax rule DV is
Applies to give nodes. According to the syntax rules in FIG. 4, the vhead rules are applied first.
In reality, this is where the past tense of give is selected, and the processing of sentence-initial adverbs and dependent clauses in complex sentences is done. next
in act head syntax rules he give himself in vg
is generated, and the object of she in <goal> is her<object
> generates book which he bought to her. During this generation, a co-occurrence relationship dictionary is used to check co-occurrence relationships (for example, <goal> does not take a preposition and the object word her connects to give).
次に、日本語文生成における共起関係辞書を用
いた処理の例を示す。 Next, an example of processing using a co-occurrence relationship dictionary in Japanese sentence generation will be shown.
第７図は文生成処理部２における意味ネツトワ
ークのアーク、すなわち２つの単語Ａ，Ｂ間の関
係指示＜object＞、＜place＞、＜goal＞などにも
とづく構文規則辞書１０からの助詞群の選択処理
例を示している。図示の例は、＜object＞につい
て、“を、が、に、について、……”が選択され
たことを示している。 FIG. 7 shows the arc of the semantic network in the sentence generation processing unit 2, that is, the formation of particle groups from the syntax rule dictionary 10 based on the relationship indications <object>, <place>, <goal>, etc. between two words A and B. An example of selection processing is shown. The illustrated example shows that "about, about,..." has been selected for <object>.
第８図は、第６図で選択された助詞群“を、
が、に、について、……”を単語Ａ，Ｂに適用す
る際に、共起関係辞書１１にある＜object＞関係
規則を参照して、最適なものを指定したところを
示している。 Figure 8 shows the particle group selected in Figure 6,
When applying "..." to words A and B, the <object> relationship rules in the co-occurrence relationship dictionary 11 are referred to and the optimal one is specified.
また、たとえば「食べる」に対して時制＜past
＞が指定されているときには、「食べるときだつ
た」となる。 Also, for example, for "eat", tense < past
When > is specified, it says "It's time to eat."
以上のようにして、本文生成処理部はまわりの
規則をとりかえるだけで各言語の質のよい文が生
成できる。 As described above, the text generation processing section can generate high-quality sentences in each language simply by changing the surrounding rules.
以上のように、本発明によれば、各言語に依存
しない自然言語文生成装置が提供できるので、各
言語ごとのプログラムを用意する必要がなくな
り、また規則の追加修正が容易となり、コストの
低減を図ることができる。
As described above, according to the present invention, it is possible to provide a natural language sentence generation device that does not depend on each language, so there is no need to prepare a program for each language, and it is easy to add and modify rules, reducing costs. can be achieved.
第１図は意味ネツトワークの説明図、第２図は
本発明の１実施例装置の構成図、第３図は英語の
単語辞書の構成例を示す説明図、第４図は構文規
則辞書の構成例を示す説明図、第５図は共起関係
辞書の構成例を示す説明図、第６図は意味ネツト
ワークの１例を示す説明図、第７図および第８図
はそれぞれ文生成処理部が構文規則辞書および共
起関係辞書を用いて行なう処理例を示す説明図で
ある。
図中、１は意味ネツトワークデータ、２は文生
成処理部、３は生成言語選択部、４は文生成規則
フアイル、５は日本語文生成規則フアイル、８は
単語辞書、９は隣接関係辞書、１０は構文規則辞
書、１１は共起関係辞書、１４は自然言語文を示
す。
FIG. 1 is an explanatory diagram of a semantic network, FIG. 2 is a configuration diagram of a device according to an embodiment of the present invention, FIG. 3 is an explanatory diagram showing an example of the configuration of an English word dictionary, and FIG. 4 is an explanatory diagram of a syntax rule dictionary. FIG. 5 is an explanatory diagram showing an example of the structure of a co-occurrence relationship dictionary. FIG. 6 is an explanatory diagram showing an example of a semantic network. FIGS. 7 and 8 are sentence generation processing, respectively. FIG. 3 is an explanatory diagram showing an example of processing performed by the unit using a syntax rule dictionary and a co-occurrence relationship dictionary. In the figure, 1 is semantic network data, 2 is a sentence generation processing unit, 3 is a generation language selection unit, 4 is a sentence generation rule file, 5 is a Japanese sentence generation rule file, 8 is a word dictionary, 9 is an adjacency relationship dictionary, 10 is a syntax rule dictionary, 11 is a co-occurrence relationship dictionary, and 14 is a natural language sentence.
Claims (1)
味ネツトワーク形式の単語情報入力手段と、文生
成規則情報フアイルと、文生成処理装置とをそな
え、上記文生成規則情報フアイルは、生成目標言
語側の単語辞書、構文規則辞書、共起関係辞書お
よび隣接関係辞書を含み、上記文生成処理装置は
入力された意味ネツトワーク形式の単語情報に含
まれる各単語について単語辞書および構成規則辞
書を参照して文章の基本的な構成範囲を設定する
とともに、共起関係辞書および隣接関係辞書にも
とづいて２つの単語間における最適な表現を選択
し文章を決定することを特徴とする自然言語文章
生成装置。1 Equipped with a word information input means in the form of a semantic network, which is a universal intermediate expression independent of language, a sentence generation rule information file, and a sentence generation processing device, and the sentence generation rule information file is used on the generation target language side. The sentence generation processing device includes a word dictionary, a syntactic rule dictionary, a co-occurrence relationship dictionary, and an adjacency relationship dictionary, and the sentence generation processing device refers to the word dictionary and the construction rule dictionary for each word included in the input word information in the semantic network format. 1. A natural language sentence generation device, which determines a sentence by setting the basic composition range of a sentence, and selecting an optimal expression between two words based on a co-occurrence relationship dictionary and an adjacency relationship dictionary.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP58182229A JPS6074081A (en) | 1983-09-30 | 1983-09-30 | Generating device for natural language sentence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP58182229A JPS6074081A (en) | 1983-09-30 | 1983-09-30 | Generating device for natural language sentence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JPS6074081A JPS6074081A (en) | 1985-04-26 |
| JPH0330183B2 true JPH0330183B2 (en) | 1991-04-26 |
Family
ID=16114598
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP58182229A Granted JPS6074081A (en) | 1983-09-30 | 1983-09-30 | Generating device for natural language sentence |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPS6074081A (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH083815B2 (en) * | 1985-10-25 | 1996-01-17 | 株式会社日立製作所 | Natural language co-occurrence relation dictionary maintenance method |
| JPH0734198B2 (en) * | 1986-06-27 | 1995-04-12 | シャープ株式会社 | Translation device |
| JPS63121977A (en) * | 1986-11-11 | 1988-05-26 | Fujitsu Ltd | Mechanical translating device |
| JP2546245B2 (en) * | 1986-11-25 | 1996-10-23 | 株式会社日立製作所 | Natural language sentence generation method |
| JPH0833895B2 (en) * | 1986-11-28 | 1996-03-29 | 富士通株式会社 | Sentence generation processing method in machine translation system |
| JP3612769B2 (en) * | 1994-05-25 | 2005-01-19 | 富士ゼロックス株式会社 | Information search apparatus and information search method |
| JP3617096B2 (en) * | 1994-05-25 | 2005-02-02 | 富士ゼロックス株式会社 | Relational expression extraction apparatus, relational expression search apparatus, relational expression extraction method, relational expression search method |
-
1983
- 1983-09-30 JP JP58182229A patent/JPS6074081A/en active Granted
Also Published As
| Publication number | Publication date |
|---|---|
| JPS6074081A (en) | 1985-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dorr | Machine translation divergences: A formal description and proposed solution | |
| Trujillo | Translation engines: techniques for machine translation | |
| Boitet et al. | Multilingual Dialogue-Based MT for monolingual authors: the LIDIA project and a first mockup | |
| Uchida | 4• 2 atlas ii: A machine translation system using conceptual structure as an interlingua | |
| Gasperin et al. | Learning when to simplify sentences for natural text simplification | |
| Boguslavsky et al. | Creating a Universal Networking Language module within an advanced NLP system | |
| JPH0330183B2 (en) | ||
| Macleod et al. | COMLEX Syntax–A Large Syntactic Dictionary for Natural Language Processing | |
| Dale 1 | A rule‐based approach to computer‐assisted copy‐editing | |
| JP2022119430A (en) | Program, device and method for inferring sentence in consideration of unknown words | |
| Mridha et al. | Development of morphological rules for bangla words for universal networking language | |
| JP3113257B2 (en) | Machine translation equipment | |
| JPH0439705B2 (en) | ||
| JPS60193074A (en) | Analyzer of japanese language | |
| Diaconescu et al. | General system for normal and phonetic inflection | |
| Yoshii | Jetr: a robust machine translation system (artificial intelligence, natural language processing, japanese) | |
| JP3353873B2 (en) | Machine translation equipment | |
| Sylvia et al. | UNL DECONVERSION FOR SENTENCE REALIZATION IN TAMIL | |
| Cho | Ku ('He'), Kyay ('S/He') and Anti-Logophoricity in Korean | |
| JP2004086919A (en) | Mechanical translation system | |
| JP3233800B2 (en) | Machine translation equipment | |
| Piotrowska et al. | Robustness in the Linguistic Automaton | |
| O'Kennon | XHOSA ON THE WEE | |
| JPH0887506A (en) | Pattern collation type translation processor | |
| Bateman | Generation of simple text structures for instructions in Bulgarian, Czech and Russian |