請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/67169
完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.advisor | 鄭卜壬(Pu-Jen Cheng) | |
dc.contributor.author | Yuh-Harn Yang | en |
dc.contributor.author | 楊喻涵 | zh_TW |
dc.date.accessioned | 2021-06-17T01:22:06Z | - |
dc.date.available | 2017-08-24 | |
dc.date.copyright | 2017-08-24 | |
dc.date.issued | 2017 | |
dc.date.submitted | 2017-08-10 | |
dc.identifier.citation | [1] Allison, J.R., Lemley, M.A., Moore, K.A., and Trunkey, R.D.: ‘Valuable patents’, George Mason Law & Economics Research Paper, 2003
[2] Grover, A., and Leskovec, J.: ‘ node2vec: Scalable feature learning for networks’, Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016 [3] Kort, F.: ‘Predicting Supreme Court decisions mathematically: a quantitative analysis of the ‘‘right to counsel’’ cases’, The American Political Science Review, 1957, 51, pp. 1-12 [4] Segal, J.A.: ‘Predicting Supreme Court cases probabilistically: the search and seizure cases, 1962–1981’, The American Political Science Review, pp. 891-900 [5] Katz, D.M., Bommarito, M.J., II , and Blackman, J.: ‘A general approach for predicting the behavior of the Supreme Court of the United States’, PLoS ONE, 2017 [6] Aletras, N., Tsarapatsanis, D., Preoμiuc-Pietro, D., and Lampos, V.: ‘Predicting judicial decisions of the European Court of Human Rights: a Natural Language Processing perspective’, PeerJ Comput. Sci., 2016 [7] Iwayama, M., Fujii, A., and Kando, N.: ‘Overview of classification subtask at NTCIR-6 patent retrieval Task’, Proceedings of the 6th NTCIR Workshop Meeting, 2007 [8] Fujii, A., Iwayama, M., and Kando, N.: ‘Overview of patent retrieval task at NTCIR-4’, In Proceedings of the 4th NTCIR Workshop, 2004 [9] Marco, A.C., Miller, R.D., Fonda, K.K., Laufer, P.M., Dzierzynski, P., and Rater, M.: ‘Patent litigation and USPTO trials: Implications for patent examination quality.’, 2016 [10] Henderson, K., Gallagher, B., Li, L., and Akoglu, L.: ‘It’s who you know: graph mining using recursive structural features’, Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining, 2011 [11] Yan, S., Xu, D., Zhang, B., Zhang, H.-j., Yang, Q., and Lin, S.: ‘Graph embedding and extensions: a general framework for dimensionality reduction’, IEEE Trans. Pattern Analysis and Machine Intelligence, 2007, pp. 40-51 [12] Mikolov, T., Chen, K., Corrado, G., and Dean, J.: ‘Efficient estimation of word representations in vector space’, ICLR Workshop, 2013 [13] Perozzi, B., Al-Rfou, R., and Skiena, S.: ‘DeepWalk: online learning of social representations’, Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining [14] Tang, J., Qu, M., Wang, M., Zhang, M., Yan, J., and Mei, Q.: ‘LINE: Large-scale Information Network Embedding.’, Proceedings of the 24th International Conference on World Wide Web, 2015 [15] Chang, C.-C., and Lin, C.-J.: ‘LIBSVM: A library for support vector machines’, ACM Transactions on Intelligent Systems and Technology, 2011, 2, (3), pp. 27:21--27:27 [16] Hochreiter, S., and Schmidhuber, J.r.: ‘Long short-term memory’, 1997, pp. 1735– 1780 [17] Kalchbrenner, N., Grefenstette, E., and Blunsom, P.: ‘A convolutional neural network for modelling sentences’, Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics., 2014, pp. 655–665 [18] Yang, Z., Yang, D., Dyer, C., He, X., Smola, A., and Hovy, E.: ‘Hierarchical attention networks for document classification’, Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2016 [19] Bahdanau, D., Cho, K., and Bengio, Y.: ‘Neural machine translation by jointly learning to align and translate’, In Procedding International Conference on Learning Representations, 2015 [20] Boser, B.E., Guyon, I.M., and Vapnik, V.N.: ‘A training algorithm for optimal margin classifiers’, In Proceedings of the Fifth Annual Workshop on Computational Learning Theory, 1992, pp. 144-152 [21] Cortes, C., and Vapnik, V.: ‘Support-vector network’, Machine Learning, 1995, pp. 273-297 [22] Johnson, R., and Zhang, T.: ‘Effective use of word order for text categorization with convolutional neural networks’, Human Language Technologies: The 2015 Annual Conference of the North American Chapter of the ACL, 2015, pp. 103-112 [23] Lai, S., Xu, L., Liu, K., and Zhao, J.: ‘Recurrent convolutional neural networks for text classification’, Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence, 2015, pp. 2267-2273 [24] Lin, R., Liu, S., Yang, M., Li, M., Zhou, M., and Li, S.: ‘Hierarchical recurrent neural network for document modeling’, Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, 2015, pp. 899-907 | |
dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/67169 | - |
dc.description.abstract | 專利是一個政府跟專利權人之間的契約。在專利權有效期間內,專利權人擁有排除他人實施其發明的權利; 作為交換,專利權人必須向社會大眾公開其發明, 以便於專利權過期之後,該發明能成為公共財產,為社會大眾所用。專利訴訟具有高度專業性與技術性,往往須花龐大的金額以及冗長的時間,除了對訴訟雙方造成龐大的經濟壓力之外,訴訟結果的不確定性也為雙方的商業運作帶來諸多不利。因此若能在訴訟啟始或具體結果出爐之前,以資訊技術準確預測專利訴訟結果, 對於商業策略的擬定將有關鍵性的影響,並且可能為雙方節省可觀的金錢與時間資源。
專利期限一般長達20年,在此期間內賦予專利權人排除他人實施發明之權利,乃是一種強大的特權授予。為了避免政府不當授予的專利權妨礙科技進展, 各國政府多制定讓社會大眾挑戰專利有效性之制度。美國專利多方複審程序是在2012年修法後新增的挑戰現行專利權之有效性的行政訴訟程序。依據美國專利與商標局截至2016年底的統計,在多方複審程序中,一旦訴願者提出之專利挑戰被成功立案, 且最終達成書面判決, 被挑戰專利至少有一個請求項被判無效的機率高達83%。因此,立案與否會造成兩造談判籌碼的顯著消長,乃是多方複審程序中一個重要的分水嶺。然而目前為止對多方複審制度立案決定的資訊分析與預測模式的研究非常的少,因此在這篇論文,我們測試了3種預測立案決定的模型, 並對這些模型進行分析比較。 我們從多方複審程序中系爭專利文件以及程序文件中抽取出三組不同的特徵作為基礎來建構預測模型。第一組是以文字為基礎的特徵,我們利用資訊獲利技術,從挑戰專利的訴願書中分離出有代表性及鑑別力的字;第二組是以網絡為基底的特徵,我們使用社交網路的概念,分析多方複審程序中有影響力的個體或關鍵字彼此之間的關係,利用機器學習得出有代表性的連續數值來表示這些個體/關鍵字的特徵值;第三組特徵主要是依循“Valuable Patent”[1]這篇在專利界影響深遠的論文中提到的「有價值專利」的特徵,從專利文件中抽取出如請求項數量,申請時間長度等特徵。 最後的結果顯示我們的模型有很不錯的準確率,做了很多實驗嘗試後,第一組特徵的準確率是最好的,ROC也可以達到優良鑑別力,表示訴願書對立案與否是關鍵,第二組特徵的ROC是三組中最好的,將網絡運用在專利關係上是很新穎又有效的特徵,第三組特徵雖然準確率並不如預期,但發現申請時間長度對立案影響很大。另外依時間序列的資料用我們的模型預測以ROC曲線分析也達到優良鑑別力,代表我們的模型在真實狀況下具有應用價值。這篇論文的其中一個目標是希望可以幫助訴願人及專利權人規劃訴訟策略,所以除了數字的結果外,我們也歸納出會預測立案的原因給他們參考。本研究是少數針對多方複審程序進行定量法律預測的前驅研究之一,可作為進一步發展複雜專利訴訟預測模型的參考。 | zh_TW |
dc.description.abstract | Patent is an agreement between the government and patent owner. At a certain time, the patent owner has the right to exclude others from implementing the invention. As an exchange, the patent owner needs to reveal the invention to public. When the right expired, it can be public property used by the community. A patent litigation is very professional and technical, so it costs huge amount of money and time. The litigation not only causes heavy economic pressures, and the uncertainty of the results also bring disadvantages for business operations of both sides. If we can predict the results precisely by information technology before the litigation begins or the results come out, it would be key effect on designing business strategies and save huge money and time for both sides.
The time of the right to exclude others for patent owner usually lasts for 20 years, so it’s very powerful. To prevent the government from giving the right improperly and obstructing progresses of technologies, there are plenty of laws for the community to challenge the rights in every country. In USA, a new law process called inter partes review(IPR) was announced by United States Patent and Trademark Office(USPTO) in 2012. As of the end of 2016 statistics by USPTO, once the IPR instituted and get the final written decision, the probability of at least one claim of the challenge patent being invalidated up to 83%. That’s why decision of institution is so important for both petitioners and patent owners, however, the previous works which study on prediction and analysis of institution decisions for IPR are very few. In this thesis, we design and analyze three predicting models for institution decision of IPR. We extracted three different kinds of features from documents of IPRs as bases to construct a predicting model. The first one is text-based features and we define a formula to find representative and discriminate terms in the petition. In graph-based features, the model uses the concept of a social network and learns continuous representation of influential entities from IPRs by the relations between each other. The last features follow “Valuable Patent”[1], a very influential paper in patent field, such as number of claims. In the end, our model can predict the decision with a strong accuracy and AUC. The accuracy of text-based feature is the best. The idea of using graph-based feature in patent prediction is novel and it get the excellent performance of AUC. Although the performance of the third feature is not so well, we find that the length of grant lag is important for decision of institution. The performance of time-series is getting better and better with time and its AUC surpasses 0.81which means our model can implement in the real world. Our aim is to help formulate strategies for both petitioners and patent owners, so we analyze the results and infer some summaries about the reasons of the prediction. Our results represent an important advance for the science of quantitative legal prediction and are practical with high commercial value. | en |
dc.description.provenance | Made available in DSpace on 2021-06-17T01:22:06Z (GMT). No. of bitstreams: 1 ntu-106-R04922061-1.pdf: 1653665 bytes, checksum: 5037d41745bc1429088e6c4cb6621bf1 (MD5) Previous issue date: 2017 | en |
dc.description.tableofcontents | 誌謝 i
摘要 iii ABSTRACT v CONTENTS vii LIST OF FIGURES xi LIST OF TABLES xiii Chapter 1 Introduction 1 1.1 Background 1 1.1.1 Definitions 1 1.1.2 Definition and Structure of Patent 1 1.1.3 Importance of the Issue: Case Study 2 1.1.4 Inter Partes Review (IPR) 3 1.1.5 Motivations & Aims 4 1.2 Problem Definitions 5 Chapter 2 Related Works 6 2.1 Litigation Predicting Model (Non-Patent) 6 2.2 Patent Analytics & Retrieval 6 2.3 Patent Litigation 7 2.4 Network Analysis 8 Chapter 3 Method 10 3.1 Overview 10 3.2 Feature (A) : Statistical Inference 11 3.2.1 Information Gain 11 3.2.2 Our Method: Discriminative Gain 12 3.2.3 Deep Learning Approach 14 3.3 Feature (B) : Entity Relation 16 3.3.1 Node2vec 17 3.3.2 Parameter Setting 21 3.4 Feature(C): Domain Knowledge 22 3.4.1 Patent-related Features 22 3.4.2 IPR-related Features 23 3.5 Model Creation 24 3.5.1 Support Vector Machine 24 3.5.2 Parameter Setting 26 3.5.3 Ensemble 27 3.6 Performance Index 28 3.6.1 Accuracy 28 3.6.2 Receiver Operating Characteristic (ROC) Curve 29 Chapter 4 Experiments and Results 30 4.1 Dataset 30 4.2 Set Prediction Performance 31 4.3 Analysis of Feature(A) : Statistical Inference 32 4.3.1 Overall Performance 32 4.3.2 Single Word 33 4.3.3 The Coverage Rate of Terms 36 4.3.4 Pseudo Count 37 4.4 Analysis of Feature(B) : Entity Relation 38 4.5 Analysis of Feature(C): Domain Knowledge 40 4.6 Feature Comparisons 41 4.6.1 Feature(A) : Statistical Inference 41 4.6.2 Feature(B) : Entity Relation 41 4.6.3 Feature(C): Domain Knowledge 42 4.7 Analysis of Models 43 4.8 Time-series Analysis 44 4.9 Summary of Effective Features 45 4.10 Case Study 47 Chapter 5 Discussions 48 5.1 Compare with “Valuable Patent” 48 5.2 Time-series Performance 48 5.3 Neural Network Approach 49 5.4 Worst Case Study 49 Chapter 6 Conclusions 51 6.1 Main Contributions 51 6.2 The Limitation and Future Work 52 REFERENCE 53 | |
dc.language.iso | en | |
dc.title | 美國專利多方複審程序立案決定之預測 | zh_TW |
dc.title | Predicting Institution Decisions of Patent Litigation: A Study on Inter Partes Review | en |
dc.type | Thesis | |
dc.date.schoolyear | 105-2 | |
dc.description.degree | 碩士 | |
dc.contributor.oralexamcommittee | 陳信希(Hsin-Hsi Chen),陳豐奇(Feng-Chi Chen),陳柏琳(Ber-Lin Chen),蔡銘峰(Ming-Feng Tsai) | |
dc.subject.keyword | 專利訴訟,自然語言處理,信息網絡,節點向量,知識特徵學習, | zh_TW |
dc.subject.keyword | Patent litigation,Natural Language Processing,Information Network,Node Embedding,Domain Knowledge Learning, | en |
dc.relation.page | 55 | |
dc.identifier.doi | 10.6342/NTU201702913 | |
dc.rights.note | 有償授權 | |
dc.date.accepted | 2017-08-10 | |
dc.contributor.author-college | 電機資訊學院 | zh_TW |
dc.contributor.author-dept | 資訊工程學研究所 | zh_TW |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-106-1.pdf 目前未授權公開取用 | 1.61 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。