JPH01140370A - Sentence paraphrasing system - Google Patents
Sentence paraphrasing systemInfo
- Publication number
- JPH01140370A JPH01140370A JP62300962A JP30096287A JPH01140370A JP H01140370 A JPH01140370 A JP H01140370A JP 62300962 A JP62300962 A JP 62300962A JP 30096287 A JP30096287 A JP 30096287A JP H01140370 A JPH01140370 A JP H01140370A
- Authority
- JP
- Japan
- Prior art keywords
- concept
- conceptual
- sentence
- conceptual structure
- dictionary
- 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.)
- Pending
Links
- 238000004458 analytical method Methods 0.000 claims description 8
- 239000000284 extract Substances 0.000 claims description 3
- 230000002457 bidirectional effect Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 6
- 238000000034 method Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 230000003796 beauty Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 241000220317 Rosa Species 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 230000000295 complement effect Effects 0.000 description 2
- 230000010365 information processing Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000003197 gene knockdown Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
Landscapes
- Machine Translation (AREA)
- Document Processing Apparatus (AREA)
Abstract
Description
【発明の詳細な説明】
（産業上の利用分野）
本発明は、情報処理分野における文章処理システムに関
する。DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a text processing system in the field of information processing.
（従来の技術）
゛情報処理技術の大衆化に伴い、文意の明確なマニュア
ル文を効率よく作成することの必要性は近年急速に高ま
っている。一方、カナ漢字変換技術、文字列編集支援技
術の進歩は、これらマニュアル文などの文書類が通常電
子的媒体上に作成されるという状況を作り出したが、こ
れらの従来技術は修正・編集・保存・流通の各面での効
率化のメリットをもたらしたのみである。現在、検索・
編集・索引作成などの機能をさらに高めた文章作成支援
システムが、［アイディア・プロセッサー」などと称し
て開発されつつあるが、概念辞書を持たず、人、力。(Prior Art) ``With the popularization of information processing technology, the need to efficiently create manual texts with clear meaning has rapidly increased in recent years. On the other hand, advances in kana-kanji conversion technology and character string editing support technology have created a situation where these manuals and other documents are usually created on electronic media, but these conventional technologies cannot be modified, edited, or stored.・It only brought about the benefits of increasing efficiency in various aspects of distribution. Currently searching/
Text creation support systems with even more advanced functions such as editing and indexing are being developed under the name of ``Idea Processors,'' but they do not have conceptual dictionaries and rely on people and power.
文章の概念構造の抽出・解析を行うことのできないこれ
らのシステムには「文意の明確さ」を正しく評価するこ
とは不可能であり、［文意の明確さの向上Ｊに直接寄与
することがない。It is impossible for these systems to extract and analyze the conceptual structure of a sentence to correctly evaluate the "clarity of the meaning of the sentence." There is no.
（発明が解決しようとする問題点）
これに対しシステムが概念辞書を持ち、入力文章の概念
構造の抽出・解析を行う手段を備えていて、さらに論理
的に互いに等価ないくつかの概念構造を認識することが
できる場合、［入力文から、基本形の概念構造への写像
のコスト」を評価関数とする等の方法によって、［文意
の明確さ」を直接に評価することが可能になる。(Problem to be solved by the invention) In order to solve this problem, the system has a concept dictionary, is equipped with means for extracting and analyzing the conceptual structure of the input text, and furthermore, it is possible to extract and analyze several conceptual structures that are logically equivalent to each other. If recognition is possible, it becomes possible to directly evaluate the ``clarity of sentence meaning'' by using a method such as ``the cost of mapping from an input sentence to a basic conceptual structure'' as an evaluation function.
具体的には、人間からみて
１）入力文で記述される概念構造には曖昧性が無いが、
否定語の多用などにより文意の理解が困難である部分、
２）概念的に不自然な関係を含んでいる部分、等が検出
できるようになる可能性がある。Specifically, from a human perspective, 1) there is no ambiguity in the conceptual structure described in the input sentence;
It may be possible to detect parts where the meaning of the sentence is difficult to understand due to excessive use of negative words, etc. 2) parts that contain conceptually unnatural relationships, etc.
裏返していえば、概念辞書及びそれを参照する前記手段
をもたない従来の文章作成支援システムには、上記１）
、２）のような部分を入力文中に検出し、さらにはより
文意の明確な出力文を生成する能力が欠けている、とい
う問題点がある。To put it another way, conventional text creation support systems that do not have a concept dictionary and the means for referring to it have the above 1).
, 2) in an input sentence, and furthermore, it lacks the ability to generate an output sentence with a clearer meaning.
（問題点を解決するための手段）
上記の問題点を解決するため本発明の文章パラフレーズ
システムは、構文解析手段と概念構造加工手段と構文生
成手段とを備え、概念見出しごとに概念素性と概念素性
型の記述をもつ概念辞書、及び概念辞書の概念見出しと
の間に双方向ポインタをもつ言語辞書の両辞書を用いて
入力文からの概念構造の抽出と概念構造からの出力文の
生成とを行う文章解析・生成システムにおいて、概念辞
書の各概念見出し情報の中に、概念素性と組み合わせて
、その概念素性に関する関連概念を関連性の型の情報と
ともに指定する、関連概念指定手段を備え、該関連概念
指定手段を利用して概念構造加工手段が概念構造のパラ
フレーズを行い、パラフレーズ結果である概念構造を構
文生成手段の入力として出力文の生成を行うという制御
を行う。(Means for Solving the Problems) In order to solve the above problems, the text paraphrase system of the present invention is provided with a syntactic analysis means, a conceptual structure processing means, and a syntax generation means, and which analyzes conceptual features for each concept heading. Extract conceptual structures from input sentences and generate output sentences from conceptual structures using both a conceptual dictionary with conceptual feature type descriptions and a linguistic dictionary with bidirectional pointers between the conceptual dictionary's concept headings. In a text analysis/generation system that performs The conceptual structure processing means paraphrases the conceptual structure using the related concept specifying means, and the conceptual structure resulting from the paraphrase is input to the syntax generating means to generate an output sentence.
（作用）
上記手段を備えることにより本発明の文章パラフレーズ
システムによれば、
１）入力文で記述される概念構造には曖昧性が無いが、
否定語の多用などにより文意の理解が困難である部分、
２）概念的に不自然な関係を含んでいる部分、を検出し
て評価し、必要に応じてより文意の明確な出力文を生成
することが可能となる。(Function) According to the text paraphrase system of the present invention by providing the above means, 1) There is no ambiguity in the conceptual structure described in the input sentence;
2) Detect and evaluate parts where the meaning of the sentence is difficult to understand due to excessive use of negative words, etc., and 2) parts that contain conceptually unnatural relationships, and output sentences with a clearer meaning if necessary. It becomes possible to generate .
（実施例）
次に本発明の実施例について、図面を参照して説明する
。第１図は本発明の一実施例である文章パラフレーズシ
ステムの構成を示すブロック図であり、第２図は概念構
造のパラフレーズの過程を示す概念図、第３図は、関連
概念指定手段の内容の例を示す図である。(Example) Next, an example of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a text paraphrase system that is an embodiment of the present invention, FIG. 2 is a conceptual diagram showing the process of paraphrasing a conceptual structure, and FIG. 3 is a related concept specifying means. It is a figure showing an example of the contents.
本実施例の関連概念指定手段２、特許請求の範囲におけ
る概念見出し情報、概念素性−関連概念、関連性の型と
いう組み合わせの情報の他に、さらに概念素性型を組み
合わせ、概念素性の主観性ｌ恒久性の度合い情報を格納
し、また、概念構造加工規則の例として［概念の否定の
規則」を格納している。The related concept specifying means 2 of this embodiment, in addition to the combination information of concept index information, concept feature-related concept, and relationship type in the claims, further combines the concept feature type, and the subjectivity l of the concept feature. It stores permanence degree information, and also stores [concept negation rules] as an example of concept structure processing rules.
以下、次の文を例に、概念構造抽出・加工・出力文生成
の過程を説明する。The process of conceptual structure extraction, processing, and output sentence generation will be explained below using the following sentence as an example.
ａ）あの人は美男ではない。a) That person is not handsome.
ｂ）美男でない人。b) A person who is not handsome.
Ｃ） ７彼は、昼というより夜に活動する。（？は不自
然な日本語を示すマーク）
構文解析手段３は、言語辞書１のみを用いて構文的手掛
かりをもとに、文節単位に概念見出しに写像させた形の
概念構造を抽出する。例えば例文ａ）では、
概念構造１：
（ＴＨＡＴ） −（指示関係）−＋（ＭＡＮ）−（属性
関係：（否定Ｈ断定））−＋（ＢＥＡＵＴＩＦＵＬ’Ｍ
ＡＮ）のようになる。C) 7 He is active at night rather than during the day. (? is a mark indicating unnatural Japanese) The syntactic analysis means 3 uses only the language dictionary 1 and extracts a conceptual structure mapped to a concept heading for each phrase based on syntactic clues. For example, in example sentence a), conceptual structure 1: (THAT) - (indicative relationship) - + (MAN) - (attribute relationship: (negative H assertion)) - + (BEAUTIFUL'M
AN).
構文解析手段３の抽出した概念構造をデ、−タ線を介し
て受は取った概念構造加工手段５は、まずその概念構造
を解析して変形（以下パラフレーズ）可能な部分を検出
する。この際の検出条件の１例としては、概念構造加工
手段５自らがもつ、数域性関係）と（否定）が共起して
、被支配概念（（ＢＥＡＵＴＩＦＵＬ’ＭＡＮ））が終
端にあるとき、（否定）は、被支配概念側に移しても、
概念構造全体の意味するところは叩じである。」という
規則と、被支配概念の見出しが関連概念指定手段２１で
ヒツトしたということの論理積があげられる。The conceptual structure processing means 5, which receives the conceptual structure extracted by the syntactic analysis means 3 via the data line, first analyzes the conceptual structure and detects parts that can be transformed (hereinafter referred to as paraphrases). An example of the detection condition in this case is when the concept structure processing means 5 itself has a number domain relation) and (negation) that co-occur, and the dominated concept ((BEAUTIFUL'MAN)) is at the end. , (negation) is transferred to the dominated concept side,
The meaning of the entire conceptual structure is a knock-down. ” and the fact that the heading of the dominated concept was hit by the related concept designation means 21.
パラフレーズ可能部分の検出（判断１）後、関連概念指
定手段２１による解析処理が起動され、まず概念の否定
の規則の規則１［単一概念の否定の焦点は、述語におけ
る否定の場合、主観性の高い順ｌ恒久性の低い順に優先
的に当たる。」が適用される。ここで第３図の［概念素
性の主観性ｌ恒久性の度合い情報」に基づく計算結果に
より
（ＢＥＡＵＴＩＦＵＬ’ＭＡＮ）の概念素性のうち、属
性１［美醜＋１に否定の焦点が付与される。After detecting a paraphrase possible part (judgment 1), the analysis process by the related concept specifying means 21 is started, and first, rule 1 of the concept negation rules [The focus of negation of a single concept is Priority is placed in the order of highest quality and lowest degree of permanence. ” applies. Here, based on the calculation result based on the [subjectivity l permanence degree information of conceptual feature] in FIG. 3, a negative focus is given to attribute 1 [Beauty/Ugliness+1] of the conceptual feature (BEAUTIFUL'MAN).
さらに規則２［否定概念の論理値は概念素性型に基づい
て求まる］を適用すると、属性耳美醜＋１の概念素性型
である［両極ｌ連続１の示す「この概念素性は直線状の
評価尺度上の十方向のある値を指し、［＋でも−でもな
い中間状態ＯＪが存在するような概念素性型である１こ
とから、その否定は［＋でも−でもない中間状態Ｏ］及
び［属性１−１という論理値を意味することとなり、下
記の概念構造解析結果が導かれる。Furthermore, by applying rule 2 [the logical value of a negative concept is found based on the concept feature type], it is a concept feature type with the attribute beauty/ugliness +1 [the bipolar l continuous 1 indicates that "this concept feature is based on the linear evaluation scale". Since it is a concept feature type 1 in which there is an intermediate state OJ that is neither + nor -, its negation is [intermediate state O that is neither + nor -] and [attribute 1- This means a logical value of 1, and the following conceptual structure analysis results are derived.
概念構造２：
（ＴＨＡＴ） −（指示関係）−＋（ＭＡＮ）−（属性
関係：（断定））−’（美醜［Ｏｏｒ −］） ａｎｄ
（性別男）ａｎｄ（人間）９次に概念構造加工手段５
では、バラフレーズ結果を出力するかバラフレーズ以前
の入力概念構造をそのまま出力するかの評価を行う。Conceptual structure 2: (THAT) - (reference relationship) - + (MAN) - (attribute relationship: (assertion)) -' (beauty and ugliness [Oor -]) and
(gender male) and (human) 9th conceptual structure processing means 5
Now, we will evaluate whether to output the rose phrase result or to output the input concept structure before the rose phrase as it is.
この評価関数は、本来概念構造全体のバランス等をも入
力パラメータとすべきものであるが、本実施例では簡単
のために「バラフレーズ時に焦点の当たっていた概念素
性（ここでは属性１［美醜＋１）。Originally, this evaluation function should also take the balance of the entire conceptual structure as an input parameter, but in this example, for the sake of simplicity, we use "the conceptual features that were focused on during the rose phrase (attribute 1 [Beauty/Ugliness + 1 ).
のパラフレーズ結果（ここでは（美醜［００ｒ−１））
に−致−する「関連性の型１をもつ関連概念が関連概念
指定手段２１中に存在するＪ場合にその部分のバラフレ
ーズ結果を出力するという関数を用いる（判断２）。Paraphrase result (here, (Beautiful and Ugly [00r-1))
A function is used that outputs a phrase result for that part when a related concept with type 1 of relevance exists in the related concept specifying means 21 (judgment 2).
関連概念指定手段２１には、概念見出し“ＢＥＡＵＴＩ
ＦＵＬ″ＭＡＮ”′の関連概念として“ＵＧＬＹ″ＭＡ
Ｎ” 、“ＢＥＡＵＴＩＦＵＬ″ＷＯＭＡＮ” 、“
旧ＥＡＵＴＩＦＵＬ”ＢＯＹ”が格納されているがこの
中でｇ性１ノｒ 関連性（７）Ｉ Ｊ ヲモ−”）”Ｕ
ＧＬＹ’ＭＡＮ”ハ、［−１という値をもっており、パ
ラフレーズ結果である［０ｏｒ−］とは一致しない。The related concept designation means 21 includes a concept heading “BEAUTI”.
“UGLY”MA is a related concept of FUL”MAN”
N”, “BEAUTIFUL”WOMAN”, “
The old EAUTIFUL "BOY" is stored, but in this G sex 1 norr Relevance (7) I J omo ") "U
GLY'MAN'' has a value of [-1, which does not match the paraphrase result [0or-].
そこで、概念構造１が構文生成手段４に送られ、出力文
が生成されて出力データ線から出力される。パラフレー
ズ部分の全く無い概念構造が構文生成手段４に与えられ
た場合でも、構文解析手段３と同じ言語辞書１が使われ
たとしても、言語辞書１から概念辞書２へのポインタと
、概念辞書２から言語辞書１へのポインタとが必ずしも
一致しないため一般には入力文と出力文とは、互いに異
なるものとなる。Therefore, the conceptual structure 1 is sent to the syntax generating means 4, and an output sentence is generated and output from the output data line. Even if a conceptual structure without any paraphrase parts is given to the syntax generating means 4, and even if the same linguistic dictionary 1 as the syntax analyzing means 3 is used, the pointer from the linguistic dictionary 1 to the conceptual dictionary 2 and the conceptual dictionary Since the pointers from 2 to the language dictionary 1 do not necessarily match, the input sentence and the output sentence are generally different from each other.
例） ｓ）−＞ ａ’）あの方は美男子だ。Example) s) -> a’) That person is a handsome man.
入力文ｂ）の構文解析手段３の出力概念構造は、′−（
属性関係）−〉”の下に（断定）が存在しない点を除い
てａ）の出力と等しい。しかし、この違いにより「概念
の否定の規則１１が適用されなくなり、概念構造加工手
段５が一旦作成する概念構造２は、ａ）の場合とは異な
ったものになる。（′人′の恒久的属性との両立性の条
件がはたらいて否定の焦点が属性３であるという可能性
は無くなるが、ｂ）のような中立的な属性表現の場合そ
の他の属性１．２に否定の焦点がある可能性は等しく残
るため、最終的な論理値はそれらの論理和１女または男
醜ｊとなる。）入力文Ｃ）を関連概念指定手段２１を用
いて概念加工手段５で解析すると、昼と夜とが互いに相
補的関係にあるにもかかわらず、（“’ＤＡＹ″ＴＩＭ
Ｅ”属性耳相補ドＮＩＧＨＴ’”［属性１“］）その両
者が結ぶＲＡＴＨＥＲ″ＴＨＡＮという概念が、属性１
［３つ以上の条件からの選択的比較を示す関係概念１を
有しているため不正な論理関係であることが検出される
（判断３）。The output conceptual structure of the parsing means 3 of the input sentence b) is '-(
It is the same as the output of a) except that (assertion) does not exist under "attribute relation) ->".However, due to this difference, "rule 11 of negation of concept is no longer applied, and the conceptual structure processing means 5 temporarily Conceptual structure 2 to be created will be different from case a) (although the possibility that the condition of compatibility with the permanent attribute of 'person' is activated and the focus of negation is attribute 3 disappears) In the case of neutral attribute expressions such as , b), there remains an equal possibility that other attributes 1 and 2 are the focus of negation, so the final logical value is their logical sum 1 female or male ugly j .) When the input sentence C) is analyzed by the concept processing means 5 using the related concept specifying means 21, even though day and night are complementary to each other, ("'DAY"TIM
E"Attribute Complementary NIGHT'" [Attribute 1"]) The concept of RATHER"THAN that connects the two is Attribute 1
[Since the relational concept 1 indicates selective comparison from three or more conditions, an invalid logical relation is detected (judgment 3).
そこで関係概念であるＲＡＴＨＥＲ″ＴＨＡＮの近似概
念を関連概念指定手段２１のなかに求めると［−３者以
上１（３者以上の関係であるという点を除き等しい概念
であるとの記号）という「関連性の型」でマークされる
“ＮＯＴ”ＢＵＴ”が存在するため、この概念で置き換
えを行う。Therefore, when an approximate concept of RATHER''THAN, which is a relational concept, is found in the related concept specifying means 21, it becomes [-3 or more 1 (symbol indicating that they are equal concepts except for the relationship of 3 or more)" Since there is a "NOT"BUT" marked with "Relationship Type", we will replace it with this concept.
この結果最終的に警告メツセージ（…り付きで彼は昼で
はなく夜に活動する。As a result of this, he finally receives a warning message (...) and is active at night instead of during the day.
という出力文を得ることが出来る。You can get the following output statement.
（発明の効呆）
以上説明したように、本発明によれば、１）入力文で記
述される概念構造には曖昧性が無いが、否定語の多用な
どにより文意の理解が困難である部分、
２）概念的に不自然な関係を含んでいる部分、を入力文
中に検出して評価し、必要に応じてより文意の明確な、
より正しい出力文を生成すること、が可能となる。(Effects of the Invention) As explained above, according to the present invention, 1) Although there is no ambiguity in the conceptual structure described in the input sentence, it is difficult to understand the meaning of the sentence due to the frequent use of negative words, etc. 2) parts that contain conceptually unnatural relationships are detected and evaluated in the input sentence, and if necessary, parts with clearer meaning of the sentence are added.
It becomes possible to generate more correct output sentences.
第１図は本発明の一実施例である文章パラフレーズシス
テムの構成を示すブロック図であり、第２図は概念構造
のパラフレーズの過程を示す概念図、第３図は、概念辞
書中の関連概念指定手段に格納された内容の例を示す図
である。
図において、Figure 1 is a block diagram showing the configuration of a text paraphrase system that is an embodiment of the present invention, Figure 2 is a conceptual diagram showing the process of paraphrasing a conceptual structure, and Figure 3 is a block diagram showing the process of paraphrasing a conceptual structure. FIG. 6 is a diagram showing an example of contents stored in a related concept specifying means. In the figure,
Claims (1)
え、概念見出しごとに概念素性と概念素性型の記述をも
つ概念辞書、及び概念辞書の概念見出しとの間に双方向
ポインタをもつ言語辞書の両辞書を用いて入力文からの
概念構造の抽出と概念構造からの出力文の生成とを行う
文章解析・生成システムにおいて、概念辞書の各概念見
出し情報の中に、概念素性と組み合わせてその概念素性
に関する関連概念を関連性の型の情報とともに指定する
関連概念指定手段を備え、該関連概念指定手段を利用し
て概念構造加工手段が概念構造のパラフレーズを行い、
パラフレーズ結果である概念構造を構文生成手段の入力
として出力文の生成を行うことを特徴とする文章パラフ
レーズシステム。A concept dictionary comprising a syntax analysis means, a concept structure processing means, and a syntax generation means, and having a description of a concept feature and a concept feature type for each concept heading, and a language dictionary having a bidirectional pointer between the concept headings of the concept dictionary. In a text analysis/generation system that extracts conceptual structure from an input sentence and generates an output sentence from the conceptual structure using both dictionaries, each concept entry information in the concept dictionary is combined with conceptual features and comprising a related concept specifying means for specifying a related concept related to a conceptual feature together with information on a type of relationship, a conceptual structure processing means paraphrasing the conceptual structure using the related concept specifying means;
A text paraphrase system characterized in that an output sentence is generated by using a conceptual structure as a paraphrase result as input to a syntax generation means.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP62300962A JPH01140370A (en) | 1987-11-27 | 1987-11-27 | Sentence paraphrasing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP62300962A JPH01140370A (en) | 1987-11-27 | 1987-11-27 | Sentence paraphrasing system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| JPH01140370A true JPH01140370A (en) | 1989-06-01 |
Family
ID=17891170
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP62300962A Pending JPH01140370A (en) | 1987-11-27 | 1987-11-27 | Sentence paraphrasing system |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPH01140370A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH03111967A (en) * | 1989-09-26 | 1991-05-13 | Nec Corp | Document editor |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS62198961A (en) * | 1986-02-27 | 1987-09-02 | Ricoh Co Ltd | Notion analyzing device |
-
1987
- 1987-11-27 JP JP62300962A patent/JPH01140370A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS62198961A (en) * | 1986-02-27 | 1987-09-02 | Ricoh Co Ltd | Notion analyzing device |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH03111967A (en) * | 1989-09-26 | 1991-05-13 | Nec Corp | Document editor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| McKeown | Paraphrasing questions using given and new information | |
| JP3220560B2 (en) | Machine translation equipment | |
| Qiu et al. | A public reference implementation of the rap anaphora resolution algorithm | |
| JPH0293866A (en) | Summary generating device | |
| JPH0510703B2 (en) | ||
| Ferrés et al. | YATS: yet another text simplifier | |
| Khan et al. | Translation divergence patterns handling in English to Urdu machine translation | |
| Sperberg-McQueen et al. | The design of the TEI encoding scheme | |
| JPH07282055A (en) | Sentence analysis method and device | |
| Muischnek et al. | Estonian dependency treebank: from constraint grammar tagset to universal dependencies | |
| Candito et al. | Enhanced UD dependencies with neutralized diathesis alternation | |
| Chen et al. | SimpleNLG-ZH: a linguistic realisation engine for Mandarin | |
| Meelen et al. | Towards a historical treebank of Middle and Early Modern Welsh, part I: Workflow and POS tagging | |
| JPH01140370A (en) | Sentence paraphrasing system | |
| Chen et al. | Yet another format of universal dependencies for Korean | |
| Rahimi et al. | Corpus based coreference resolution for Farsi text | |
| Spencer | Factorizing lexical relatedness | |
| El-Taher et al. | An Arabic CCG approach for determining constituent types from Arabic Treebank | |
| JP2870279B2 (en) | Translation selection device | |
| Mohanty et al. | Semantically relatable sets: building blocks for representing semantics | |
| Weitz | Improving retrieval performance of translation memories using morphosyntactic analyses and generalized suffix arrays | |
| Isahara et al. | Japanese-English aligned bilingual corpora | |
| Marcoux et al. | Formal and informal meaning from documents through skeleton sentences | |
| JP2915113B2 (en) | Natural language automatic translation system | |
| Del Grosso et al. | Structuring Arabic lexical and morphological resources using TEI: theory and practice |