プレス発表文 (11 例目) 自動識別型 画像フィルタリングシステム
まず原文から。 (こちらでも原文と添削例を対照できます。)
E-Guardian Inc. and University of Tokyo Started Collaboration to Research “Auto Picture Filtering Systems” Technology
E-Guardian Inc. (Tokyo) and the University of Tokyo started collaborating on research in the field of "Auto Picture Filtering Systems" in order to advance the big date analysis for enhanced internet safety
========
E-Guardian Inc. [listed on Stock Exchange of Tokyo Mothers](http://www.e-guardian.co.jp/ head office,Minato-ku, Tokyo CEO：Takatani)have started an industry-academic collaboration for the purpose of researching the technology called “Auto Picture Filtering Systems” together with department of Mechano-Informatics at Graduate School of Information Science & Technology of the National University of Tokyo,(Main campus：Bunkyo-ku, Tokyo, the university president is Junichi Harada, the program lead is Professor Tatsuya Harada, (http://www.mi.t.u-tokyo.ac.jp/)
These days, the number of pictures and movies posted by both users and companies is increasing rapidly as the use of internet media, especially smart phones with cameras, has risen dramatically.
Likewise, the number of picture and movie platform services have become popular, driving up the amount of distribution channels facilitating these services.
Additionally, in Japan people are growing more and more interested in the safety of the web content itself. For example, child pornography has become an important object of public concern. In fact the Japanese government is drafting the new law (bill submitted to parliament No. 183)which is to state that every internet provider has responsibility to for protect children from sexual exploitation on the internet.
However, unlike text, the ability to monitor pictures and movies for its contents is currently not well supported by systems. As such there is a need for monitoring each image by manpower. Monitoring by manpower is of course accurate but it drives up cost.
Therefore, there are high expectations for the"Auto Picture Filtering Systems"technology because it both enables the ability to scale up the monitoring of numerous of data content as well as reduce cost.
As such, E-Guardian which aims to be worldwide internet patrolling company and the University of Tokyo plan to collaborate leveraging E-Guardian’s many years of experience in monitoring and huge amount of ‘training data’(1)as well as the world leading “Picture recognition Systems” developed by Harada’s research laboratory at University of Tokyo.
(The training data is to be learned as 'practice data', which is used to develop the recognition and monitoring systems. During the training sessions the system is double checked visually by manned operation so amendments can be made accordingly.)
The plan is that E-Guardian develops these types of custom made filtering systems tailored to the requirements of companies and government institutes, and continuously improves the technology further over time leveraging the industry-academic collaboration. It is projected at this stage that the outcome of the research leads to at least 2/3 cost reduction while yielding the same results compared with 100% visual monitoring by manpower.
=================
全体として、説明の流れ自体はわかりやすく、個々の文に文法上の誤りも少ないが、肝心の「何を開発するのか」「まったく新規の開発なのか、それとも既存技術を応用したシステムを開発するのか」「(大学との) 新たな共同作業なのか、それとも既存の協力の強化なのか」 という点が曖昧であり、説得力に欠けている。要点を簡潔に言う努力を避け、むしろ冗長な表現で要点をぼかしているようにさえ思える。以下は原文を再掲しながらのコメント。最初は表題と副題から。
E-Guardian Inc. and University of Tokyo Started Collaboration to Research “Auto Picture Filtering Systems” Technology
E-Guardian Inc. (Tokyo) and the University of Tokyo started collaborating on research in the field of "Auto Picture Filtering Systems" in order to advance the big date analysis for enhanced internet safety
「東京大学」の英語表記は The University of Tokyo なのだが、プレス発表で正式名を使わないのは致命的誤り。Started Collaboration to Research は「共同研究」のつもりかもしれないが、全体を読むと「共同研究」ではなく、既存の研究技術を基に新たなシステムの開発を始めたようなので、そのことを明確に言うべきだろう。また、Filtering Systems と複数にしているが、実際に開発するシステムは 1 つと思えるので、単数の方が適切のはず。さらに、Filtering Systems Technology と言っているが、「技術」を開発するのではなく、「システム」を開発するはずなので、Technology は不要。
サブ見出しの advance the big date analysis は「分析を前に進める」と読めてしまう。また internet safety と「安全性」を言っているが、実際にはコンテンツが適切なものかどうかを判断するシステムのようなので、safety よりも legitimacy などを使った方が本来は適切と思う。ただし見出しとしては internet safety の方が適切かもしれない。internet は先頭大文字で Internet とした方が無難と思う。
続いて本文。
E-Guardian Inc. [listed on Stock Exchange of Tokyo Mothers](http://www.e-guardian.co.jp/ head office,Minato-ku, Tokyo CEO：Takatani)have started an industry-academic collaboration for the purpose of researching the technology called “Auto Picture Filtering Systems” together with department of Mechano-Informatics at Graduate School of Information Science & Technology of the National University of Tokyo,(Main campus：Bunkyo-ku, Tokyo, the university president is Junichi Harada, the program lead is Professor Tatsuya Harada, (http://www.mi.t.u-tokyo.ac.jp/)
最も重要な最初の一文に「上場」や URL、所在地などを入れた理由が理解できない。これらは発表文の最後に About ... のようにまとめるのが普通であり、その方がはるかに記事も読みやすくなる。こうした修飾部分が多すぎ、しかも for the purpose of など冗長な表現を使っている (for のみで十分) ため、肝心の発表内容が非常に読み取りにくい。「東京大学」の英語表記が不適切なことも問題だが、表題での表記とも違っているため、読む人は必ずしも同じ「東京大学」と認識してくれないおそれもある。
These days, the number of pictures and movies posted by both users and companies is increasing rapidly as the use of internet media, especially smart phones with cameras, has risen dramatically.
内容を読めば最近のことを言っているのは自明なので These days は不要。user は individuals (個人) のつもりだろうか。誰が投稿した内容なのかを後から取り上げるのでない限り、誰が投稿したかは重要ではないはずなので、この部分も不要。後半は use ... has risen という文になっているが、use が rise することはあり得ないのでは。
Likewise, the number of picture and movie platform services have become popular, driving up the amount of distribution channels facilitating these services.
主語と述語の関係が number... have become popular となっており、変である。その次も、channel が可算名詞であるにもかかわらず amount of ... channels となっている。
Additionally, in Japan people are growing more and more interested in the safety of the web content itself. For example, child pornography has become an important object of public concern. In fact the Japanese government is drafting the new law (bill submitted to parliament No. 183)which is to state that every internet provider has responsibility to for protect children from sexual exploitation on the internet.
ネットのコンテンツが適切かどうか懸念するのは日本人のみではないので、in Japan は不要と思う。interested は「関心・興味を持つ」であり、「懸念」とは違う。ここでは、上で書いたように safety を legitimacy などにすべきだろう。has become an important object of public concern は has become a major public concern などとすれば簡潔で読みやすくなる。日本の「国会」には parliament を使わず、diet を使うのが原則のはず。
However, unlike text, the ability to monitor pictures and movies for its contents is currently not well supported by systems. As such there is a need for monitoring each image by manpower. Monitoring by manpower is of course accurate but it drives up cost.
unlike text で始めたら、pictures and movies ... と続けるのが自然。「能力がサポートされていない」という書き方もおかしい。there is a need では「需要がある」という意味になってしまう。
Therefore, there are high expectations for the"Auto Picture Filtering Systems"technology because it both enables the ability to scale up the monitoring of numerous of data content as well as reduce cost.
「期待が高まっている」とでも言いたいらしいが、それならば because や both を使うのではなく、technology that will scale up the data monitoring and reduce the cost at the same time などとすべきだろう。
As such, E-Guardian which aims to be worldwide internet patrolling company and the University of Tokyo plan to collaborate leveraging E-Guardian’s many years of experience in monitoring and huge amount of ‘training data’(1)as well as the world leading “Picture recognition Systems” developed by Harada’s research laboratory at University of Tokyo.
(The training data is to be learned as 'practice data', which is used to develop the recognition and monitoring systems. During the training sessions the system is double checked visually by manned operation so amendments can be made accordingly.)
この部分は本来、「そのため当社は・・・をします」と具体的な内容を明示する部分のはずだが、この原文では肝心の要点がよくわからない。仕方がないので添削例は「既存の世界的技術を基に、共同でシステムを開発する」という趣旨の文にしてある。原文の ( ) で囲まれた段落は「当社データを使用してシステムを学習させる」と言っているのだと思うが、添削例ではその前の段落と一緒にまとめてある。
The plan is that E-Guardian develops these types of custom made filtering systems tailored to the requirements of companies and government institutes, and continuously improves the technology further over time leveraging the industry-academic collaboration. It is projected at this stage that the outcome of the research leads to at least 2/3 cost reduction while yielding the same results compared with 100% visual monitoring by manpower.
この部分も肝心な点が意味不明で困る。原文は「顧客ごとにカスタマイズしたシステムを開発する」と読めるが、それでは顧客ごとにいくつも別システムを開発することになってしまう。添削例では、「カスタマイズ可能な 1 つのシステムを開発する」という趣旨にしてある。原文で言う「少なくとも 2/3 のコストダウン」は「現状よりも 2/3 下がる」のか「現状の 2/3 に下がるのか」曖昧だが、添削例では「すべて目視で監視する場合よりも少なくとも 60% 下がる」としてある。
=================
以下は添削例
E-Guardian Inc. and The University of Tokyo jointly develop Auto Picture Filtering System
E-Guardian Inc. and The University of Tokyo collaborate on Auto Picture Filtering System for big data analysis and Internet safety
========
E-Guard Inc. started a project with a department of The University of Tokyo on an automatic picture filtering system for the Web. The project is part of the company's collaboration with academics in the field of Web technologies. The department, Mechano-Informatics at Graduate School of Information Science & Technology, has been developing the picture filtering technology under the lead professor, Tatsuya Harada (http://www.mi.t.u-tokyo.ac.jp/).
With a drastic increase of Internet devices, especially smart phones with camera, the number of pictures and movies posted on the Web is increasing exponentially. Similarly, the number of platform services for pictures and movies is increasing, broadening distribution channels.
At the same time, people are getting more aware of the legitimacy of the Web content. As an example, child pornography is becoming a major public concern. The Japanese government is drafting a new law submitted to the 183rd diet session that makes every Internet provider responsible for protecting children from sexual exploitation on the Internet.
Unfortunately, unlike texts, pictures and movies are not easy for current systems to adequately watch, and thus need to be manually watched. The manual work is more accurate, but costs more.
There is a strong need for an automatic filtering system for pictures and movies. As it will be an automatic system, it can watch significantly larger amount of data than the current level and drastically reduce cost at the same time.
With the goal of being a worldwide Internet patrolling company, E-Guardian leverages its many years of experience in Web monitoring and huge amount of data obtained and will build a new filtering system through a collaboration with The University of Tokyo. The new system will be based on the world-leading “picture recognition system" developed by the professor Harada's lab and will be fed the company's data for "training." While the system undergoes the "training," operators manually check the results to improve the accuracy of the system.
E-Guardian plans to develop the system to be easily customizable to each requirement of companies and/or government institutions through ongoing cooperation with academics including the professor Yamada's lab. The system is expected to reduce the content-monitoring cost by at least 60% compared with all-manual operation while yielding the same result.
コメント 0