JP2018077604A - Artificial intelligence device automatically identifying violation candidate of achieving means or method from function description - Google Patents
Artificial intelligence device automatically identifying violation candidate of achieving means or method from function description Download PDFInfo
- Publication number
- JP2018077604A JP2018077604A JP2016217807A JP2016217807A JP2018077604A JP 2018077604 A JP2018077604 A JP 2018077604A JP 2016217807 A JP2016217807 A JP 2016217807A JP 2016217807 A JP2016217807 A JP 2016217807A JP 2018077604 A JP2018077604 A JP 2018077604A
- Authority
- JP
- Japan
- Prior art keywords
- technology
- product
- noun
- phrase
- service
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Machine Translation (AREA)
Abstract
Description
本発明は、作成された文章内容について自然言語処理を行い、その記述文章を構成する内容が類似する文章をデータベースから抽出し、類似する要点を抽出し、類似する内容部分と相違する点の分析を行う人工知能を実現するシステムおよび人工知能を実現するソフトウエアに適用して有効な技術に関する The present invention performs natural language processing on the prepared sentence content, extracts sentences with similar contents constituting the description sentence from the database, extracts similar points, and analyzes points that are different from similar contents parts Technology that is effective when applied to systems that implement artificial intelligence and software that implements artificial intelligence
従来の自然言語処理系の人工知能において、文章の要点がどこであるかの特定方法として、（１）文章から抽出した単語の頻出度による重要度の分析で要点部分を特定する方法、（２）いわゆるディープラーニングにより抽出した名詞同士を関連付けてデータベース化し、新規の文章で同じ名詞が出た場合にデータベース化した関連付けた名詞が含まれるか否かで判断する手法である。 （１）の場合、文章の書き方など文章作者の書き方に依存性が高く、頻出度が異なるため、要点部分の特定は困難であることが既知である。 （２）は、すでに確定しているデータベースに対して、同様な単語が出現した場合に確率は上がるが、あくまでも名詞同士の関連付けによるので、同様な記述の文章が出現するかあるいは、類似度の高い文章が多く格納されるまでは、精度が上がらないという問題点があることが既知である。
そのため、技術を使った文章中の技術・手法を抽出し、その技術・手法の類似点や相違点を他の文章か自動抽出する精度をあげることは、困難であった。さらに文章を解読しているわけではないため、文章の差異を自動抽出することは、従来困難であった。
In the conventional artificial intelligence of a natural language processing system, as a method for identifying the point of a sentence, (1) a method of identifying a main part by analyzing the importance based on the frequency of words extracted from the sentence, (2) This is a technique in which nouns extracted by so-called deep learning are associated with each other to form a database, and when the same noun is generated in a new sentence, it is determined whether or not the associated nouns included in the database are included. In the case of (1), it is known that it is difficult to specify the main part because it is highly dependent on the writing method of the writing author, such as how to write the writing, and the frequency of occurrence is different. In (2), the probability rises when similar words appear in a database that has already been confirmed, but because it is only based on the association of nouns, a sentence with the same description appears, It is known that there is a problem that accuracy does not increase until many high sentences are stored.
For this reason, it has been difficult to extract techniques / methods in sentences using technology and automatically extract similarities and differences of the techniques / methods from other sentences. Furthermore, since the sentence is not decoded, it has been difficult in the past to automatically extract the difference between sentences.
自然言語処理による人工知能において、製品やサービスの内容、製品やサービスに使用する技術、製品に使用する部品内容のそれぞれの記述から自然言語処理によって、候補から構成技術・手法を特定し、その技術・手法を実現する手段・方法が同一の出願済み知財権、権利確定済知財権、公知技術あるいは任意の技術・手法を侵害可能性として自動特定し、回避方法を自動作成する In artificial intelligence based on natural language processing, component technologies and methods are identified from candidates by natural language processing based on descriptions of products and services, technologies used for products and services, and parts used for products.・ Applied IP rights, vested IP rights, known technologies, or any technology / method with the same means / method for realizing the method are automatically identified as infringement and an avoidance method is automatically created
本願の発明の概略を説明すれば、以下のとおりである。すなわち本発明は、製品やサービスの内容、製品やサービスに使用する技術、製品に使用する部品内容のそれぞれの記述文章に対して、自然言語処理を行い、その結果をもとに文節ごとに分け、文節同士の関係性、およびそれぞれの文節内の名詞および名詞に係る品詞、および動詞から作成した動名詞の組み合わせを用い、実現する技術・手法候補を特定し、その技術・手法候補を構成する手段・方法の構成を侵害可能性候補として自動特定し、手段・方法を構成する異なる項目を組み合わせることで、回避策を自動作成する。 The outline of the invention of the present application will be described as follows. In other words, the present invention performs natural language processing on the description sentences of the contents of products and services, the technology used for products and services, and the contents of parts used for products, and divides them into phrases based on the results. Identify the technologies / method candidates to be realized using the combination of clauses, the nouns in each clause, the part of speech associated with the noun, and the combination of verb nouns created from the verbs, and configure the technologies / method candidates The configuration of the means / method is automatically specified as a potential infringement candidate, and the workaround is automatically created by combining different items constituting the means / method.
以下、本発明の実施の形態を図面に基づいて詳細に説明する。ただし、本発明は多くの異なる態様で実施することが可能であり、本実施の形態の記載内容に限定して解釈すべきではない。なお、実施の形態の全体を通して同じ要素には同じ番号を付するものとする。 Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. However, the present invention can be implemented in many different modes and should not be interpreted as being limited to the description of the present embodiment. Note that the same numbers are assigned to the same elements throughout the embodiment.
以下の実施の形態では、主に方法またはシステムについて説明するが、当業者であれば明らかなとおり、本発明はコンピュータで使用可能なプログラムとしても実施できる。したがって、本発明は、ハードウェアとしての実施形態、ソフトウェアとしての実施形態またはソフトウェアとハードウェアとの組合せの実施形態をとることができる。プログラムは、ハードディスク、ＣＤ−ＲＯＭ、光記憶装置または磁気記憶装置、半導体記憶装置等の任意のコンピュータ可読媒体に記録できる。 In the following embodiment, a method or system will be mainly described. However, as will be apparent to those skilled in the art, the present invention can also be implemented as a program usable on a computer. Therefore, the present invention can take an embodiment as hardware, an embodiment as software, or an embodiment of a combination of software and hardware. The program can be recorded on any computer-readable medium such as a hard disk, a CD-ROM, an optical storage device or a magnetic storage device, or a semiconductor storage device.
また以下の実施の形態では、一般的なコンピュータシステムを用いることができる。実施の形態で用いることができるコンピュータシステムは、中央演算処理装置（ＣＰＵ）、グラフィックス プロセッシング ユニット（ＧＰＵ）、主記憶装置（メインメモリ：ＲＡＭ）、不揮発性記憶装置（ＲＯＭ）、コプロセッサ、画像アクセラレータ、キャッシュメモリ、入出力制御装置（Ｉ／Ｏ）等、一般的にコンピュータシステムに備えられるハードウェア資源を備える。また、ハードディスク装置等の外部記憶装置、インターネット等のネットワークに接続可能な通信手段を備えることができる。コンピュータシステムには、パーソナルコンピュータ、ワークステーション、メインフレームコンピュータ等各種のコンピュータやタブレットコンピュータ、スマートフォン等各種携帯情報端末が含まれる。 In the following embodiments, a general computer system can be used. A computer system that can be used in the embodiment includes a central processing unit (CPU), a graphics processing unit (GPU), a main memory (main memory: RAM), a nonvolatile memory (ROM), a coprocessor, an image Generally, hardware resources such as an accelerator, a cache memory, and an input / output control device (I / O) are provided in a computer system. In addition, a communication unit that can be connected to an external storage device such as a hard disk device or a network such as the Internet can be provided. The computer system includes various computers such as personal computers, workstations, and mainframe computers, and various portable information terminals such as tablet computers and smartphones.
本発明は、自然言語処理をコアとする人工知能において、任意の文章に含まれる、製品やサービスの内容、製品やサービスに使用する機能・技術、製品に使用する部品に使用される機能・技術の手法・手段をそれぞれに記述される説明の内容について自然言語処理によって、課題と実現する技術を実現する技術・手法を特定し、その技術を実現する手段・方法が出願済み知財権、権利確定済知財権、公知技術あるいは任意の技術・手法から特定することにある。 The present invention relates to contents of products and services, functions and technologies used for products and services, functions and technologies used for parts used in products, in artificial intelligence with natural language processing as a core. The contents of the explanation that describes each of the methods and means are identified by the natural language processing to identify the technology and technique that realizes the problem and the technology to be realized. The purpose is to specify from fixed IP rights, publicly known technologies, or arbitrary technologies / methods.
図１は、本発明の実施の形態である自然言語処理をコアとする人工知能構成の一例を示した概念図である。図１の機能・技術推論（００１）、手段・方法処理（００２）を含む。 FIG. 1 is a conceptual diagram showing an example of an artificial intelligence configuration having natural language processing as a core according to an embodiment of the present invention. 1 includes function / technical reasoning (001) and means / method processing (002) in FIG.
機能・技術推論は、製品やサービスの内容、製品やサービスの機能・技術仕様の文章（テキスト文）を公知技術である、ＭｅＣａｂ，ＣａｂｏＣｈａ等に代表される形態素解析、係り受け解析を用いる自然言語処理を行い、機能・技術内容を抽出・特定し、その抽出結果を比較対象となる文章（群）を同様自然言語処理を用いて抽出した機能・技術内容と比較し、類似した候補を導き出す。ここでの文章とは、音声テキスト変換後の文章、キーボード等の入力装置からの文章を含む。手段・方法処理では、機能・技術推論の類似候補を機能・技術の実現手段・方法を、公知技術である、ＭｅＣａｂ，ＣａｂｏＣｈａ等に代表される形態素解析、係り受け解析を用いる自然言語処理を行い、比較対象となる文章（群）をから抽出する。抽出結果を実現手段・方法候補をする。 Function / technical reasoning is a natural language that uses morphological analysis and dependency analysis represented by MeCab, CaboCha, etc., which are well-known techniques for the contents of products and services, and texts (text sentences) of the functions and technical specifications of products and services. Processing is performed to extract / specify the function / technical content, and the extracted result is compared with the function / technical content extracted using the same natural language processing for the sentence (group) to be compared, and similar candidates are derived. The sentences here include sentences after voice text conversion and sentences from an input device such as a keyboard. In the means / method processing, similar candidates for function / technical reasoning are processed by means of natural language processing using means / methods for realizing functions / techniques, morphological analysis represented by MeCab, CaboCha, etc., which are known techniques. Then, the sentence (group) to be compared is extracted from. The extraction result is made a realization means / method candidate.
図２は、機能・技術推論、手段・方法の処理の流れのフロー図である。
（１）製品やサービスの内容、製品やサービスに使用する機能・技術、製品に使用する部品に使用される機能・技術内容を記述される説明の文章（音声変換からテキスト変換したものを含む）を文節に分ける。文節の分解（文節抽出）は、句読点での分割を基本とし、分割した句読点間（文の始まりから最初の句読点までの場合も含む）に動詞を含む場合は、次の句読点までを文節する。動詞を含まない場合は、次の句読点までに動詞を含むかどうかを確認し、動詞を含むまで繰り返し、その繰り返した句読点間（文の始まりから最初の句読点までの場合も含む）すべてを１つの文節とする。
（２）次に元の文章を形態素解析、係り受け解析を使用して、品詞レベルの単語に分け、単語同士の係り受け関係を得る。（１）で分けた文節毎にこの係り受け関係をあてはめ、文章として、係り受けの単語が一番多く掛かってくる最終位置の名詞または動詞を含む文節を主文節とする。主文節以外を従文節とする。従文節は、主文節が実現する機能・技術に対する構成要素項目となる。
（３）動詞は、文節に分けた後、動名詞として扱う。
（４）文節分解後に、それぞれの文節内で重み付けを行う。文節内で係り受け解析を行い、単語同士の係る関係性を係り受け解析の結果によってと、文節内の関係性を得る。文節内の最後にくるのが名詞の場合は、この名詞を起点として、文節内で名詞・動名詞を係りかたで係る段数が近いほうから密の関係、離れるほど疎の関係とする。文節内で最後にくるのが、動詞（動名詞）の場合は、直近の名詞句を起点として、文節内での名詞・動名詞係りかたで近いほうから密の関係、離れるほど疎の関係とする。同時に従文節は、動名詞化した動詞にかかる名詞と、その名詞に係る品詞の関係から技術・手法の構成する１つの項目として、その文節における名詞、名詞に係る品詞（修飾）および動名詞の組み合わせという関係を得る。（文節処理）
（５）（４）の結果を用い、各従文節が名詞のみまたは、名詞と名詞に係る品詞、動名詞の組み合わせを粗密の関係を用いて比較対象となる文章、文章群、データベース内の文章群（出願済み知財権、権利確定済知財権、公知技術の実施例）から（１）〜（４）の同じの処理によって得た結果とを比較して、類似性を比較する。全一致が完全一致であり、全一致しない場合は、全項目から疎の関係の名詞の順に比較対象から外して比較する。以下不一致の場合は、疎の名詞を外しながら比較を行う。（文節比較） あるいは密の関係からその関係の順に比較項目を加えていく。これを類似度として判断する。以下不一致の場合は、疎の名詞を加えながら比較を行う。（文節比較）
この時に文節内の名詞、動名詞の関係は接続詞、助詞によって、AND、OR、XOR、NOTの関係を抽出し、すべてAND置き換えて比較する。（例 A AND (B OR C)の場合は、A AND BとA AND Cに分割し、係る関係でそれぞれに対して疎密の関係を作成し、それぞれ比較する）
（1）によって作成された文節に含まれる動詞が「備える」、「所持する」「有する」「形成する」をはじめとする所持、条件を関連付けさせる動詞のみの場合は、この動詞を動名詞化し、この動名詞と、それにかかる名詞を、技術・手法を構成する項目の構成要素としての比較項目に含める。（1）によって作成された文節に含まれる動詞が「備える」、「所持する」「有する」「形成する」をはじめとする所持、条件を関連付けさせる動詞以外の動詞をもつ場合は、所持、条件を関連付ける動詞を含めずに、他の動詞を動名詞化し、この動名詞と、それにかかる名詞を、技術・手法を構成する項目の構成要素としての比較項目に含める。
（６）一定の割合以上（任意）の類似度を持った文節をもつ文章を、元の文章が表す機能・技術内容が一致する候補と判断する。
（７）次に機能・技術内容が一致する候補の文章から実現手段・方法候補を抽出する。候補の文章の（出願済み知財権、権利確定済知財権、公知技術の実施例）から手段、実施形態を対象として、文節に分ける。文節の分解（文節抽出）は、句読点での分割を基本とし、分割した句読点間（文の始まりから最初の句読点までの場合も含む）に動詞を含む場合は、次の句読点までを文節する。動詞を含まない場合は、次の句読点までに動詞を含むかどうかを確認し、動詞を含むまで繰り返し、その繰り返した句読点間（文の始まりから最初の句読点までの場合も含む）すべてを１つの文節とする。
（８）次に元の文章を形態素解析、係り受け解析を使用して、品詞レベルの単語に分け、単語同士の係り受け関係を得る。（１）で分けた文節毎にこの係り受け関係をあてはめ、文章として、係り受けの単語が一番多く掛かってくる最終位置の名詞または動詞を含む文節を主文節とする。主文節以外を従文節とする。従文節は、主文節が実現する手段・方法に対する構成要素項目となる。
（９）動詞は、文節に分けた後、動名詞として扱う。
（１０）文節分解後に、それぞれの文節内で重み付けを行う。文節内で係り受け解析を行い、単語同士の係る関係性を係り受け解析の結果によってと、文節内の関係性を得る。文節内の最後にくるのが名詞の場合は、この名詞を起点として、文節内で名詞・動名詞を係りかたで係る段数が近いほうから密の関係、離れるほど疎の関係とする。文節内で最後にくるのが、動詞（動名詞）の場合は、直近の名詞句を起点として、文節内での名詞・動名詞係りかたで近いほうから密の関係、離れるほど疎の関係とする。同時に従文節は、動名詞化した動詞にかかる名詞と、その名詞に係る品詞の関係から技術・手法の構成する１つの項目として、その文節における名詞、名詞に係る品詞（修飾）および動名詞の組み合わせという関係を得る。（文節処理）
（１１）（１０）の結果を用い、各従文節が名詞のみまたは、名詞と名詞に係る品詞、動名詞の組み合わせを粗密の関係を用いて比較対象となる文章、文章群、データベース内の文章群と類似性を比較する。全一致が完全一致であり、全一致しない場合は、全項目から疎の関係の名詞の順に比較対象から外して比較する。以下不一致の場合は、疎の名詞を外しながら比較を行う。（文節比較） あるいは密の関係からその関係の順に比較項目を加えていく。これを類似度として判断する。以下不一致の場合は、疎の名詞を加えながら比較を行う。（文節比較）
この時に文節内の名詞、動名詞の関係は接続詞、助詞によって、AND、OR、XOR、NOTの関係を抽出し、すべてAND置き換えて比較する。（例 A AND (B OR C)の場合は、A AND BとA AND Cに分割し、係る関係でそれぞれに対して疎密の関係を作成し、それぞれ比較する） （７）によって作成された文節に含まれる動詞が「備える」、「所持する」「有する」「形成する」をはじめとする所持、条件を関連付けさせる動詞のみの場合は、この動詞を動名詞化し、この動名詞と、それにかかる名詞を、技術・手法を構成する項目の構成要素となる。（７）によって作成された文節に含まれる動詞が「備える」、「所持する」「有する」「形成する」をはじめとする所持、条件を関連付けさせる動詞以外の動詞をもつ場合は、所持、条件を関連付ける動詞を含めずに、他の動詞を動名詞化し、この動名詞と、それにかかる名詞を、技術・手法を構成する項目の構成要素となる。
（１２）これらの構成要素を再接続処理したものが、手段・方法を構成する１つの項目であり、その文節のもととなる文章に応じた文節の組み合わせで手段・方法を構成する１つの項目を再接続処理したものが、出願済み知財権、権利確定済知財権、公知技術の手段・方法を抽出したものとなり、他者の権利化済みである場合、侵害の可能性候補となる。
（１３）侵害可能性の手段・方法が存在する場合、回避方法の自動作成は、（Ａ）手段・方法を構成する項目の正負論理の組み合わせ、（ｉＢ）物質・物性・サイズの置き換え、 （Ｃ）手段・方法を構成する他の手段・方法の項目に置き換えとなる。（図３）
（１４）手段・方法を構成するそれぞれの文節から得る項目の正負論理の組み合わせは、従文節を手段・方法の構成要素として、（１０）による名詞と動名詞に係る名詞句の組み合わせに対する負論理を作成する。それぞれの文節に対する負論理を作成する。この負論理の項目（負論理の手段・方法の項目）と正論理の文節からなる項目（正論理の手段・方法の項目）について、手段・方法を構成する項目をすべて組み合わせる。ここで、すべて正論理であるオリジナルと対となる正負の手段・方法の項目は組み合わせは、作成しない。（元の文の構成する項目が、それぞれ、Ａ、Ｂ、Ｃとする場合、Ａ＆Ｂ＆Ｃ、Ａ＆Ａ−（負論理）のような組み合わせは作成しない）。これを手段・方法の正負組み合わせ候補とする。この手段・方法の正負組み合わせ候補との比較対象となるデータベース、文章（群）に対して、（７）〜（１２）と同様に文節抽出、文節処理、文節比較によって文節ごとから抽出した名詞とそれにかかる品詞、動名詞の組み合わせ、文節をつなぐ接続詞による組み合わせた結果との比較をする。 比較の結果、手段・方法の正負組み合わせ候補の中で、比較一致しない組み合わせが、回避方法の自動作成結果となる。
（１５）物質・物性・サイズの置き換えは、（７）〜（１０）の変換処理後の文節抽出において名詞に係る品詞の抽出において、任意の物質・物性・サイズに関する名詞を抽出した場合は、任意の物質・物性・サイズに関する名詞を含む文節から作成した項目すべてを対象に物質・物性は、異なる物質・物性に置き換えて、サイズは、与えられたサイズ以外に置き換えた組み合わせを作成する。その結果を比較対象となるデータベース、文章（群）に対して、（７）〜（１２）と同様に文節抽出、文節処理、文節比較によって文節ごとから抽出した名詞とそれにかかる品詞、動名詞の組み合わせ、文節をつなぐ接続詞による組み合わせた結果との比較をする。 比較の結果、手段・方法の項目の組み合わせ候補の中で、比較一致しない組み合わせが、回避方法の自動作成結果となる。
（１６）手段・方法を構成する他の手段・方法の項目に置き換えは、（７）〜（１０）による文節から得る手段・方法の各項目によって実現する結論に一致または類似する結論を導き出す他の手段・方法の各項目に置き換える。（７）〜（１０）による文節分解による手段・方法の項目についての名詞とそれにかかる品詞、動名詞の組み合わせを作成。次にこの項目を使用し比較対象となるデータベース、文章（群）に対して（７）〜（１０）と同様に文節抽出、文節処理、文節比較によって抽出した名詞とそれにかかる品詞、動名詞の組み合わせを比較をする。 比較の結果、項目に一致した手段・方法から（７）〜（１０）の処理により目的を主文節から確定する。目的に一致する他の手段・方法を比較対象となるデータベース、文章（群）に対して（７）〜（１２）と同様に文節抽出、文節処理、文節比較によって抽出した名詞とそれにかかる品詞、動名詞の組み合わせを対象を主文節に比較をする。不一致の場合は、候補が存在しないと判断し終了する。一致する場合、この文章に対して、 （７）、（８）と同様の処理を行い得た従文節を、元の比較対象である文節と置き換える。これが回避方法の自動作成結果である。
FIG. 2 is a flowchart of the process flow of function / technical reasoning and means / method.
(1) Contents of products and services, functions and technologies used for products and services, and explanatory texts that describe functions and technologies used for parts used in products (including those converted from speech to text) Is divided into clauses. Phrase decomposition (phrase extraction) is based on division at punctuation, and if a verb is included between the divided punctuation marks (including the case from the beginning of the sentence to the first punctuation mark), the next punctuation is punctuated. If it does not contain a verb, check to see if it contains a verb by the next punctuation, repeat until it contains a verb, and repeat all of the repeated punctuation (including from the beginning of the sentence to the first punctuation). It shall be a phrase.
(2) Next, the original sentence is divided into words of part-of-speech level using morphological analysis and dependency analysis, and the dependency relationship between words is obtained. This dependency relationship is applied to each of the clauses divided in (1), and the clause containing the noun or verb at the final position where the dependency word is most frequently applied is used as the main clause. The sub clauses other than the main clause are used. The sub clause is a component item for the function / technology realized by the main clause.
(3) Verbs are treated as verbal nouns after being divided into phrases.
(4) After the phrase decomposition, weighting is performed in each phrase. Dependency analysis is performed within a phrase, and the relationship between words is obtained based on the result of the dependency analysis. When a noun comes at the end of a phrase, the noun and verbal noun are used in the phrase as a starting point. When a verb (verb) comes last in a phrase, the closest noun / verb nouns in the phrase start from the nearest noun phrase, the closer the closer, the more sparse the further And At the same time, the subclause is a noun in the phrase, the part of speech (modification) of the noun, the part of the noun in the phrase, and the part of the verb noun. Get the relationship of combination. (Phrase processing)
(5) Using the results of (4), each subclause is a noun only, or a sentence, sentence group, or sentence in the database to be compared using a close relationship between a noun and part of speech related to a noun and verbal noun. The similarities are compared by comparing the results obtained from the same processing of (1) to (4) from the group (filed IP rights, vested IP rights, examples of known technology). When all matches are complete matches and not all matches, comparison is performed by removing all items from the comparison target in the order of sparsely related nouns. In the case of disagreement below, comparison is performed while removing sparse nouns. (Phrase comparison) Or add comparison items in the order of the relationship from the close relationship. This is determined as the similarity. In the case of disagreement below, a comparison is made while adding sparse nouns. (Phrase comparison)
At this time, the relationship between nouns and verbal nouns in the clause is extracted by AND and OR, XOR and NOT by conjunctions and particles, and all are compared by AND replacement. (Example: In the case of A AND (B OR C), divide into A AND B and A AND C, and create a sparse / dense relationship for each of them and compare them)
If the verb included in the clause created in (1) is the only verb possessing, including “having”, “having”, “having”, “forming”, and the verbs that associate conditions, this verb is converted to a verbal noun This verbal noun and its related noun are included in the comparison item as a component of the item constituting the technology / method. If the verb included in the clause created in (1) has a verb other than the possession, possession, possession, and formation, and other verbs that associate the condition, possession, condition Other verbs are converted into verb nouns, without including the verbs that relate to, and the nouns and their nouns are included in the comparison items as components of the items constituting the technology / method.
(6) A sentence having a phrase having a certain degree or more (arbitrary) similarity is determined as a candidate having a matching function / technical content represented by the original sentence.
(7) Next, extraction means / method candidates are extracted from candidate sentences having the same function / technical content. The candidate sentences (applied intellectual property rights, vested intellectual property rights, examples of known technology) are divided into clauses for the means and embodiments. Phrase decomposition (phrase extraction) is based on division at punctuation, and if a verb is included between the divided punctuation marks (including the case from the beginning of the sentence to the first punctuation mark), the next punctuation is punctuated. If it does not contain a verb, check to see if it contains a verb by the next punctuation, repeat until it contains a verb, and repeat all of the repeated punctuation (including from the beginning of the sentence to the first punctuation). It shall be a phrase.
(8) Next, the original sentence is divided into words of part-of-speech level using morphological analysis and dependency analysis, and the dependency relationship between words is obtained. This dependency relationship is applied to each of the clauses divided in (1), and the clause containing the noun or verb at the final position where the dependency word is most frequently applied is used as the main clause. The sub clauses other than the main clause are used. The sub clause is a component item for the means / method realized by the main clause.
(9) Verbs are treated as verbal nouns after being divided into phrases.
(10) After the phrase decomposition, weighting is performed in each phrase. Dependency analysis is performed within a phrase, and the relationship between words is obtained based on the result of the dependency analysis. When a noun comes at the end of a phrase, the noun and verbal noun are used in the phrase as a starting point. When a verb (verb) comes last in a phrase, the closest noun / verb nouns in the phrase start from the nearest noun phrase, the closer the closer, the more sparse the further And At the same time, the subclause is a noun in the phrase, the part of speech (modification) of the noun, the part of the noun in the phrase, and the part of the verb noun. Get the relationship of combination. (Phrase processing)
(11) Using the results of (10), each subclause is a noun only, or a sentence, sentence group, or sentence in a database to be compared using a close relationship between a noun and part of speech related to the noun and verbal noun. Compare group and similarity. When all matches are complete matches and not all matches, comparison is performed by removing all items from the comparison target in the order of sparsely related nouns. In the case of disagreement below, comparison is performed while removing sparse nouns. (Phrase comparison) Or add comparison items in the order of the relationship from the close relationship. This is determined as the similarity. In the case of disagreement below, a comparison is made while adding sparse nouns. (Phrase comparison)
At this time, the relationship between nouns and verbal nouns in the clause is extracted by AND and OR, XOR and NOT by conjunctions and particles, and all are compared by AND replacement. (Example: In the case of A AND (B OR C), divide into A AND B and A AND C, create a sparse / dense relationship for each of them, and compare each) The clause created by (7) If the verb contained in the phrase is only possessed, possessed, possessed, possessed, and other verbs that associate conditions, this verb is converted into a noun, and this verb noun is associated with it Nouns are a component of items that make up technologies and methods. If the verb included in the clause created in (7) has a verb other than the possession, possession, possession, form, and other verbs that associate the condition, possession, condition Other verbs are converted into nouns, without including the verbs that relate to, and these nouns and their nouns are the constituent elements of the technology and method.
(12) A reconnection process of these components is one item constituting the means / method, and one means / method is constituted by a combination of clauses corresponding to the sentence that is the basis of the clause. If the item is reconnected, it will be the extracted intellectual property rights, vested intellectual property rights, and methods / methods of known technology. Become.
(13) When there is a means / method of infringement possibility, automatic creation of the avoidance method includes (A) a combination of positive and negative logic of items constituting the means / method, (iB) replacement of substance / physical property / size, C) It is replaced with the item of another means / method constituting the means / method. (Figure 3)
(14) The combination of the positive and negative logics of items obtained from the respective clauses constituting the means / method is the negative logic for the combination of the noun phrase related to the noun and the verbal noun according to (10), with the subordinate clause as the constituent element of the means / method. Create Create negative logic for each clause. For this negative logic item (negative logic means / method item) and an item consisting of positive logic clauses (positive logic means / method item), all the items constituting the means / method are combined. Here, a combination of positive and negative means / methods paired with the original, which is all positive logic, is not created. (If the items constituting the original sentence are A, B, and C, combinations such as A & B & C and A & A- (negative logic) are not created). This is a positive / negative combination candidate of means / method. For the database and sentence (group) to be compared with the positive / negative combination candidate of this means / method, the noun extracted from each sentence by phrase extraction, phrase processing, and phrase comparison as in (7) to (12) Compare the results with the combination of part of speech, verbal noun combination, and conjunctions connecting phrases. As a result of the comparison, among the positive / negative combination candidates of the means / method, the combination that does not compare and match becomes the automatic creation result of the avoidance method.
(15) Substitution of substance / physical property / size is as follows: when extracting a noun related to any substance / physical property / size in the extraction of the part of speech related to the noun in the phrase extraction after the conversion processing of (7) to (10), For all items created from clauses including nouns related to any substance / physical property / size, the substance / physical property is replaced with a different material / physical property, and the size is replaced with a combination other than the given size. For the database and sentence (group) to be compared, the nouns extracted from each phrase by phrase extraction, phrase processing, and phrase comparison, and the part of speech and verbal nouns as in (7) to (12) Compare with the combined results of conjunctions that connect combinations and clauses. As a result of the comparison, among combinations of means / method items, combinations that do not compare and match are automatically created avoidance methods.
(16) Substitution with other means / method items constituting the means / method is to derive a conclusion that agrees with or is similar to the conclusion realized by each item of the means / method obtained from the clauses according to (7) to (10) Replace with each item of method / method. Create a combination of nouns, part-of-speech and verbal nouns for items of means / methods based on clause decomposition according to (7) to (10). Next, using this item, nouns extracted by phrase extraction, phrase processing, and phrase comparison, and parts of speech and verbal nouns, as in (7) to (10), for the database and sentence (group) to be compared Compare combinations. As a result of the comparison, the purpose is determined from the main phrase by the processing of (7) to (10) from the means / method that matches the item. Nouns extracted by phrase extraction, phrase processing, phrase comparison as in (7) to (12) for the database, sentence (group) to be compared with other means and methods that match the purpose, and the part of speech associated with the noun Compare verbal noun combinations to the main sentence. If they do not match, it is determined that no candidate exists and the process ends. If they match, the subordinate clauses obtained by performing the same processing as (7) and (8) on this sentence are replaced with the original comparison target clause. This is an automatic creation result of the avoidance method.
特許などの知財権の調査において、出願前考案や出願済み知財権に対して、パテントトロール対策として、パテントトロールが取得する可能性のある技術・手法について、事前に候補を作成し、危険性の調査・警告するシステムに利用 In a search for intellectual property rights such as patents, candidates for potential technologies and techniques that patent trawls may acquire are prepared in advance as countermeasures against patent trawls for pre-application and patented IP rights. Used in systems to investigate and alert
Claims (2)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016217807A JP2018077604A (en) | 2016-11-08 | 2016-11-08 | Artificial intelligence device automatically identifying violation candidate of achieving means or method from function description |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016217807A JP2018077604A (en) | 2016-11-08 | 2016-11-08 | Artificial intelligence device automatically identifying violation candidate of achieving means or method from function description |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| JP2018077604A true JP2018077604A (en) | 2018-05-17 |
Family
ID=62150364
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP2016217807A Pending JP2018077604A (en) | 2016-11-08 | 2016-11-08 | Artificial intelligence device automatically identifying violation candidate of achieving means or method from function description |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JP2018077604A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102051835B1 (en) * | 2019-03-29 | 2019-12-04 | 김영상 | A selling system with assembling parts according to the order |
| KR20200103176A (en) * | 2019-02-12 | 2020-09-02 | 이용규 | 3D printing modeling-based intellectual property service drive system, 3D printing modeling intellectual property service driven app for the same |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002358407A (en) * | 2001-06-01 | 2002-12-13 | Sonzaisha:Kk | Selection system for business model |
| JP2003216623A (en) * | 2002-01-17 | 2003-07-31 | Ricoh Co Ltd | Document analysis device and method and recording medium |
| JP2003281186A (en) * | 2001-11-13 | 2003-10-03 | Posco | Example base retrieval method and retrieval system for determining similarity |
| JP2009238131A (en) * | 2008-03-28 | 2009-10-15 | Nomura Research Institute Ltd | Author comparison system |
| JP2010224623A (en) * | 2009-03-19 | 2010-10-07 | Nomura Research Institute Ltd | Method and program for recommending related article |
-
2016
- 2016-11-08 JP JP2016217807A patent/JP2018077604A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002358407A (en) * | 2001-06-01 | 2002-12-13 | Sonzaisha:Kk | Selection system for business model |
| JP2003281186A (en) * | 2001-11-13 | 2003-10-03 | Posco | Example base retrieval method and retrieval system for determining similarity |
| JP2003216623A (en) * | 2002-01-17 | 2003-07-31 | Ricoh Co Ltd | Document analysis device and method and recording medium |
| JP2009238131A (en) * | 2008-03-28 | 2009-10-15 | Nomura Research Institute Ltd | Author comparison system |
| JP2010224623A (en) * | 2009-03-19 | 2010-10-07 | Nomura Research Institute Ltd | Method and program for recommending related article |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200103176A (en) * | 2019-02-12 | 2020-09-02 | 이용규 | 3D printing modeling-based intellectual property service drive system, 3D printing modeling intellectual property service driven app for the same |
| KR102200700B1 (en) * | 2019-02-12 | 2021-01-11 | 이용규 | 3D printing modeling-based intellectual property service drive system, 3D printing modeling intellectual property service driven app for the same |
| KR102051835B1 (en) * | 2019-03-29 | 2019-12-04 | 김영상 | A selling system with assembling parts according to the order |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109241538B (en) | Chinese entity relation extraction method based on dependency of keywords and verbs | |
| Nguyen et al. | Event detection and domain adaptation with convolutional neural networks | |
| US9690772B2 (en) | Category and term polarity mutual annotation for aspect-based sentiment analysis | |
| US8762132B2 (en) | Textual entailment recognition apparatus, textual entailment recognition method, and computer-readable recording medium | |
| US11334608B2 (en) | Method and system for key phrase extraction and generation from text | |
| JP7100747B2 (en) | Training data generation method and equipment | |
| JP2020126493A (en) | Paginal translation processing method and paginal translation processing program | |
| Agrawal et al. | Affective representations for sarcasm detection | |
| US10853569B2 (en) | Construction of a lexicon for a selected context | |
| JP2019082931A (en) | Retrieval device, similarity calculation method, and program | |
| Qian et al. | A transition-based model for joint segmentation, pos-tagging and normalization | |
| Fabregat et al. | Extending a Deep Learning Approach for Negation Cues Detection in Spanish. | |
| Nehar et al. | An efficient stemming for arabic text classification | |
| JP6867963B2 (en) | Summary Evaluation device, method, program, and storage medium | |
| Nooralahzadeh et al. | Part of speech tagging for french social media data | |
| Chennoufi et al. | Impact of morphological analysis and a large training corpus on the performances of Arabic diacritization | |
| JP2018077604A (en) | Artificial intelligence device automatically identifying violation candidate of achieving means or method from function description | |
| JP5911931B2 (en) | Predicate term structure extraction device, method, program, and computer-readable recording medium | |
| Ali et al. | Detection of plagiarism in Urdu text documents | |
| JP2018073298A (en) | Method for automatic extraction/creation of means/method by artificial intelligence device | |
| US20230140938A1 (en) | Sentence data analysis information generation device using ontology, sentence data analysis information generation method, and sentence data analysis information generation program | |
| US20210073258A1 (en) | Information processing apparatus and non-transitory computer readable medium | |
| Lee et al. | Syllable-based Malay word stemmer | |
| Makrynioti et al. | Sentiment extraction from tweets: multilingual challenges | |
| JP6067616B2 (en) | Utterance generation method learning device, utterance generation method selection device, utterance generation method learning method, utterance generation method selection method, program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| A621 | Written request for application examination |
Free format text: JAPANESE INTERMEDIATE CODE: A621 Effective date: 20180315 |
|
| A977 | Report on retrieval |
Free format text: JAPANESE INTERMEDIATE CODE: A971007 Effective date: 20190308 |
|
| A131 | Notification of reasons for refusal |
Free format text: JAPANESE INTERMEDIATE CODE: A131 Effective date: 20190402 |
|
| A02 | Decision of refusal |
Free format text: JAPANESE INTERMEDIATE CODE: A02 Effective date: 20191029 |