基於分解機器之社群影響力分析研究－以GitHub為例 - 政大學術集成

全文

(1)˙º⌃„_hK>§qˇõ⌃êv Â GitHub ∫ã Social Influence Analysis based on Factorization Machines: Using GitHub as an Example  v ⇢5ÛI Student⇢Chih-Wei Chang ⌥ Yà⇢!òZÎ ãÁ9ZÎ Advisor⇢Dr. Ming-Feng Tsai, Dr. Chuan-Ju Wang. 立. ↵À?ª'x 治 «政 ⌦—x˚. 大. ©Î÷á. ‧. ‧ 國. 學 A Thesis. n. er. io. sit. y. Nat. submitted to Department of Computer Science National Chengchi University in partial afulfillment of the Requirements iv l. C. forhthe i U e ndegree g c h of Master in. n. Computer Science. -Ô⌘↵. ~ˆm t ]. September 2017.

(2) 記錄編號：G0104753039. 國立政治大學博碩士論文全文上網授權書 National ChengChi University Letter of Authorization for Theses and Dissertations Full Text Upload (提供授權人裝訂於紙本論文書名頁之次頁用) (Bind with paper copy thesis/dissertation following the title page) 本授權書所授權之論文為授權人在國立政治大學資訊科學學系系所 ________________組 106學年度第一學期取得碩士學位之論文。 This form attests that the _____________ Division of the Department of Graduate Institute of Computer Science at National ChengChi University has received a Master degree thesis/dissertation by the undersigned in the _________ semester of 106 academic year. 論文題目（Title）：基於分解機器之社群影響力分析研究－以 GitHub 為例 ( Social Influence Analysis based on Factorization Machines: Using GitHub as an Example ) 指導教授（Supervisor）：蔡銘峰,王釧茹立書人同意非專屬、無償授權國立政治大學，將上列論文全文資料以數位化等各種方式重製後收錄於資料庫，透過單機、網際網路、無線網路或其他公開傳輸方式提供用戶進行線上檢索、瀏覽、下載、傳輸及列印。國立政治大學並得以再授權第三人進行上述之行為。. The undersigned grants non-exclusive and gratis authorization to National ChengChi University, to reproduce the above thesis/dissertation full text material via digitalization or any other way, and to store it in the database for users to access online search, browse, download, transmit and print via singlemachine, the Internet, wireless Internet or other public methods. National ChengChi University is entitled to reauthorize a third party to perform the above actions. 論文全文上載網路公開之時間（Time of Thesis/Dissertation Full Text Uploading for Internet Access）：網際網路（The Internet） ■ 立即公開 ● 立書人擔保本著作為立書人所創作之著作，有權依本授權書內容進行各項授權，且未侵害任何第三人之智慧財產權。 The undersigned guarantees that this work is the original work of the undersigned, and is therefore eligible to grant various authorizations according to this letter of authorization, and does not infringe any intellectual property right of any third party. ● 依據96年9月22日96學年度第1學期第1次教務會議決議，畢業論文既經考試委員評定完成，並已繳交至圖書館，應視為本校之檔案，不得再行抽換。關於授權事項亦採一經授權不得變更之原則辦理。 According to the resolution of the first Academic Affairs Meeting of the first semester on September 22nd, 2007,Once the thesis/dissertation is passed after the officiating examiner's evaluation and sent to the library, it will be considered as the library's record, thereby changing and replacing of the record is disallowed. For the matter of authorization, once the authorization is granted to the library, any further alteration is disallowed，立書人：張至偉簽名(Signature)：中華民國年月日 Date of signature：__________/__________/__________ (dd/mm/yyyy) . gsweb.cgi.html[9/15/17, 12:51:35].

(3) Ù. ,÷á/(. + !ò Yàå ãÁ9 YàÑâ√⌥. ã +Ñ_á≤↵∫⌘S↵xxÑ˙ ì⌘ | vÑ±≈ ˚q↵✏≤↵Gì⌘ cãÀÜ„ÇUÙ fÑÕ\˝õÙ⌦. d◆. 立 itÑ©ÎÌ. ÕBB…. ‹. Nat. !ƒ |åEÊÍ⌘⇥ v!Å óÚÀ xw. ≈Y. FR( SocTeam Ñ. io. √. *ÜÑÂP⌘_⌥Â. s◊. ì⌘˝Î. ⌥k©⇥. +⌘∫. SocTeam Ñv%4. xwÑ. ⌘ FM Ñ¿ı. xw⌥. ⌘xSÂX⌥vπ. ≤. å·vOLÑ8 Û/Ñ. Ù§. ⌥v. er. #FR⇥. Ùæ2Ü⌘(«ô. y% ! + º–fi œ ˆ∫ãi⇥iM +x. ⇣ÑK¶. Ùì⌘x“0⇡[Ö∫ÑÕÅ'. º˚. ‧. ‧ 國. ’. ; ì⌘Ù˝⇣üÑb. ↵Ñ©Y. 學. —xZvÑ˝õ ⌘Ö∫UãÑS⌃ XıZ. ì⌘(© 治 ‘˚ 政大 ©åBÑ WSM ≤↵. +·˚⌘Ñ˝õ. «⌦—x Há(˚. y. !. sit. f⇥. ↵å⇣Ñ. ÷( CLIP ÊW§· w™õÑ%4 1œ ↵´®Ñ'∂≠ ˝ w ÷OL _˝ w⌃´√≈⌦Ñú✓¿⇥⇥ ⇥⌫Ñ≠V⌥e. n. al. ˜. Ch. iM(Â\AŸB. ↵ ≤ -. engchi. i Un. Ωz⌥Y⌘ñ∫. åh üâÑ∞K ˝. v. ≤ŸH˛. ì⌘û. ( åH˛˙Æ<Ñ⇣ú⇥. û'x0©ÎÌ¯GÑ¥∫ P ôKáåÚU⌃⇥(Ä ©ÎÑ Ô⌦ 1⇢Ñ+ò ⇡ M xÑí¯v å◆ı⇥ Ká⌥U⌃( WSM ≤↵Ñ 0√™. ì⌘b. OL˝Œ⇤. ÷áá’Ñ⌥. =/. ≠viÑ. P( CLIP Ñj4. „⇥. w. (ÊW§·® xxl✏ Ñ Î↵✏Î0ÿ⇠BìÑX(⇥ M↵À(⇡ Ô⌦Ñ¯: ä@ õÊÑÂPä⇣ „ÿÑfi∂⇥ å. ⌘Ñ8Ω⌥⌘Ñ∂∫. ⌘⇥. ⇡µ©Î. Ñ+w. ∂∫. x. Ø@. ∂∫. ˛fà/. Ñ↵À. ⌘Ñ. 9Âd«÷á{f⌘. Â œ ↵k©N⌘Ñ↵À⌘. ↵À?ª'x«⌦—x˚ September 2017. 1. ·˚⌘=˘ `⌘⇥ 5ÛI.

(4) ˙º⌃„_hK>§qˇõ⌃êv. Â GitHub ∫ã. -áXÅ >§T \sÑ˙˛u Ü .Ñ∞Ñ \b✏ ÔÂì∫⌘ /⇠‘(T \Ñs⌦ ÂÊ˛q Ó⇡∫ÓÑ 2L(6K ìÑí’åã|⇥Â GitHub fl‘T \s∫ã ( Hã|Ñ \N↵- ds⇠⌅Ü@ √⌥(⇧Ñí’N↵ ⇡õí’N↵ ⌅W(⇧ º H¢{Ñ↵¶K _±+W(⇧|dKìÑ qˇ˝õ⇥,á… ÷⌥aœqˇõÜí N↵-Ñ¢{⇧ Â√2 (6KìÙ ∆Ñ \‹¬⇥ ≥qÑ˙º bÑπ’(U⌃dÔLB ⇤‡∫ bh:’Ñ∑ P „Âä(6åT \ HÑM ⌦o å-«ô Metadata t 0˙À!ãÑN↵- !’åt·(6( \Ñ HKìÑí ’N↵⇥‡d ,v–˙ .)(®¶!ã Üt (6⌥ \ HÑí’N↵ &(x“ÑN↵-†e(6⌥ H↵✏º-Ñ API « ⌦ Ü!Ït↵T \N↵-ÑqˇõÙc≥^Ñ≈b⇥✏Nd! ã ,÷á–˙Ñπ’⌥ÔÂ,œœ↵(6 T \ HÑ[(q ˇõ< 2 aœ˙œ (6 ºt↵>§qˇõ Âû GitHub ê∆ Ñ Êx⁄∆⌦2LÊW I ,v@–˙π’K H' (˙º õ≤Ô⌦–õÑí ˙ñ ,÷á–˙Ñπ’ÔÂ–õÙ}Ñqˇ õí Pú⇥ddK Âñ∫ Ñπ✏H˛ÊWÑPú û-¿fl ˙↵✏ºÑ API «⌦ ºœ GitHub Ñ>§qˇõÑÕÅ'⇥. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 2. i Un. v.

(5) Social Influence Analysis based on Factorization Machines: Using GitHub as an Example. Abstract The emergence of community collaboration platform creates a new form of cooperation that allows people or groups to interact and develop a project with users for the purpose of achieving common goals. Taking GitHub, a collaboration platform, as an example, this platform records the detailed interaction of all participating users in the process of project development. This paper aims to discuss and measure the influence of the contributors using their interactions and the additional information of projects on the platform. In specific, this study proposes a framework to integrate the interaction between users and collaborative projects and, in the process, to learn to merge the user and the project code in the API information so as to simulate the entire process of cooperation under the impact of the proliferation of transmission of user influence. The proposed method is able to measure the potential impact of each user on collaborative projects and thus the impact of each user on the entire community of GitHub collected from the real dataset in the experiments. The experimental results show that the proposed method provides better ranking results than several baseline methods. In addition, this thesis provides a visualization of the experimental results.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 3. i Un. v.

(6) Ó⌅ Ù. 1. -áXÅ. 2. Abstract , ‡ “÷ . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 M . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 vÓÑ . . . . . . . . . . . . . . . . . . . . . . . ,å‡ ¯‹á{¢ . . . . . . . . . . . . . . . . . . . . . . 2.1 >§≤ÔÑ⌃ê . . . . . . . . . . . . . . . . . . . 2.2 ®¶!ãó’ . . . . . . . . . . . . . . . . . . . 2.2.1 T N˛ Collabrative Filtering . . . . 2.2.2 ˙ºgπÑN˛ Content-based Filtering 2.2.3 ˜ ãó’ Hybrid Algorithm . . . , ‡ vπ’. . . . . . . . . . . . . . . . . . . . . . . . . 3.1 ®¶˚q⌥>⇤T \≤ÔÑ^‘ . . . . . . . 3.2 [(qˇõÈcÑI€N↵ . . . . . . . . . . . . 3.3 ↵∫>§qˇõÑœ ˝✏ . . . . . . . . . . . . ,€‡ ÊWPú⌥ ÷ . . . . . . . . . . . . . . . . . . . . 4.1 «ô∆ . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 ↵✏º©⌦o . . . . . . . . . . . . . . . 4.2 ÊW-ö . . . . . . . . . . . . . . . . . . . . . . . 4.3 U0⌥⇡ . . . . . . . . . . . . . . . . . . . . . . . 4.4 ÊWPú . . . . . . . . . . . . . . . . . . . . . . . 4.5 ñ∫ . . . . . . . . . . . . . . . . . . . . . . . . 4.5.1 í ¡Í↵MÑ˛a⌃ê . . . . . . . . . ,î‡ P÷ . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 P÷ . . . . . . . . . . . . . . . . . . . . . . . . . .. 立. 政治大. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. n. al. y. er. io. sit. Nat. . . . . . . . . . . . . . . .. ‧. ‧ 國. 學. . . . . . . .. Ch. engchi. 4. i Un. v. . . . . . . . . . . . . . . . . . . . . . . .. 3 1 1 1 3 3 4 4 5 5 6 6 7 9 12 12 12 14 14 15 17 20 21 21.

(7) Ó⌅ 3.1 >§T \≤Ô- [ ¢{⇧ - ¢{⇧ ] Èc⌥ [ ¢{⇧ ] Èc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 [(qˇõÈcÑI€N↵ . . . . . . . . . . . . . . . . . 3.3 Â⌃„_hœ ↵∫>§qˇõ⌃xÑN↵ . . . . . . . . . 3.4 ⌃„_hÑ8e<✏ƒã . . . . . . . . . . . . . . . . . . . .. 7 8 9 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 17 18 19 19. 政治 .大 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ‧. ‧ 國. 學. 4.1 ñ∫ ≤Ÿ∂À . . . . . . . . . . . 4.2 Â∞ÎNÑ H=xíèÑ⇣,Pú 4.3 FM !†eM «⌦Ñ⇣,Pú . . . . 4.4 FM †eM «⌦Ñ⇣,Pú . . . .. 立. H . . . . . . . . . . . .. n. er. io. sit. y. Nat. al. Ch. engchi. 5. i Un. v.

(8) hÓ⌅ h 4.1 Git Awards ≤Ÿgí M30 Ñ Python ã|∫· h 4.2 M30 Ñ⇣,PúU0⌃x . . . . . . . . . . . . h 4.3 20 !ÊWU0⌃xÑsG<⌥äpx . . . . . . h 4.4 M10 Python modules/packages Ñ†⌦⌦Õ . . . h 4.5 (6 vinta ¡ Ñ HÑq «⌦ . . . . . . . .. 立. 政治大. . . . .. Æ⌥q . . . . . . . . . . . . . . . . . . . .. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 6. i Un. v. «ô . . . . . . . . . . . . . . . .. 13 15 15 16 20.

(9) ,. ‡. “÷ 1.1. M. 政治大. 立. ‧ 國. ⌃´|dÑÛ’. ÂX. Repository. X2´. ÂÊ˛. ↵q. ÑÓ⇡⇥q. „z¯. OLÜÊ˛⇥. 學. ®W>§≤Ô Social Network åT \s Collaborative Social Network ãÇ0GitHub å Quora Ñ˙˛ ∫⌘ãÀ(>§≤Ô-⌥÷∫q Ê\ H ÑÓ⇡⇢8✏Nq User. Ñ(6. å. Ê\ H. Nat. ≤Ô-(6Ñqˇõ≥≠Ñπ✏ Ù⇢Ü„ [2, 3, 4]⇥. io. —tÜ>§≤ÔÑv-. |˛Ü˝. œ. sit. y. \Ñ’Kä å. (⇧Ñqˇõ&∫ö⇡.. al. er. T. ‧. Project KìÑí’g ±+ (6Ñqˇõ⌦o ⇡#Ñ⌦oÔ˝⇤ T \≤Ô-Ñv÷∫ ⇣u'Ñqˇ⇥⌃êÇdÑ>§qˇõ ÔÂ⌘⌘ q. \. v Ñ0π (º>§T i n Ch \≤ÔÑâ 0(6 Bã|⇢↵ H ⇡õU H˝„hÜ≤Ô-(6Ñq e n gchi Ó⇡⇥ãÇ GitHub /ã|∫·|d§í&(–õX2´⌦T Â\Ñs⇥ ‹uÑ√⌥⇧ÑÕÅ' [7, 8]⇥⌥≥q>§≤Ô. n. ≤Ô-. ⇢N⇡õX2´. ã|∫·⌥v¢{ÑË/–§f. ⇣Ü ↵˙º HÑT. HX2´Ñ¡. ⇧. û. À. \≤Ô⇥. 1.2  v ÓÑ ∫Ü. H0UI⇡õ±+º≤ÔÑ(6. ↵π’. ✏N⌥>§T. ⌘(≥qÑ˙º(6⌥. HÑqˇõ⌦o. (d÷á-–˙Ü. \≤Ô⌥®¶˚q^‘Üœ. (6Ñqˇõ⇥⌘. HÑ®¶˚q. ⌥˙º¢{⇧⌥. Hí’‹¬Ñ>§. T \sKì2L^‘⇥(,á-–˙ÑF∂↵ T N˛ Filtering ⇡.´„€(º®¶˚qÑÇı (fö ↵>§T Çı ºœ. ≤Ô-(⇧qˇõÑπb WÕÅÑ\(⇥. 1. Collaborative \≤Ô d.

(10) Factorization Machines; FM. ⌘⌘✏N_h⌃„. [13]. ◆Ù!ã⌥,œ. í’‹¬Ñ[(qˇõ<⇥y%Ñ0π(º ‡∫ FM -yµÂ↵ Feature Engineering Ñ ,' Generality åH;' Flexibility ì⌘⌘ÔÂ⌥ GitHub >§≤Ôg(⇧Ñ↵✏º⌦o N↵ œ œ↵¢{⇧⌥. Âyµ⌘œÑπ✏. e FM !ãÑx“. HKìÑ‹¬qˇõ⇥. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 2. i Un. v.

(11) ,å‡ ¯‹á{¢ 2.1. 政治大. >§≤ÔÑ⌃ê. 立. ¢. 學. ‧ 國. ,áÑv/ˆåÂ DBLP «ô—xxS÷áÑ«ô∆. >§⇠ñÑqˇõ. Ñœ v [16] vÑÓ⇡ˆå(>§-(⇧ÑT\í’‹¬ >§qˇõ(>§s⌦Ñí’N↵⇥. Ñ|U. @ÂÇU⌃êåœ. v>§≤Ôå>§qˇõÑ‹u˚Ÿ⇥‡d. io. >§≤Ô-Ñqˇõ. (⇡. ⇠flÚì. Ëqˇåì•qˇ [8, 12, 14]⇥. n. al. Ç DBLP å /. à⇢. .^. er. Ñv Ç;Lqˇ. \s. e¢. y. I>§sÎ. åT. sit. Nat. GitHub. ‧. ãÇ Facebook å Twitter. ®W>§≤Ÿ. 2. Ch. i Un. v. ⌦-⌅^ãÑv ⌃êÜ∫⌘ÇU(>§≤Ô-¯íqˇ Â ≤Ô⌅↵⇣·ÑqˇõÇU(>⇤≤Ô-≥≠⇥€Âq™ (>§T\Ñ≤Ô- Ô Âv∫⌘|dí¯qˇÑ‹¬ ⇧⇥. º>§≤Ÿ. engchi. &…d~˙¢{⇤⇢. ãÇ Facebook å Twitter. /⇤. qˇõÑ¢{. (6àÔ˝⇤(>§≤Ô-. æ⌘˝®⇤ qˇõÑ↵À ⇧ªfi ’ö’ f⇤ qˇõÑ(6Ñ «á‡ G⇥⌘⌘ç∫⇡/ ↵;LdbÑqˇõ ⇡✏sW(6úa⇡↵;L⇥ Lu Liu et al. 2012. [8] –˙Ü. ↵_á. ⇣!ã. vÑÓÑ(º⇣,(6/&⇤úa–↵;L. Üœ. ;LdbÑqˇõ⇥⇡⇧. &˝. œ. ˙>§≤Ô-–. ↵;LÑqˇõ⇥d rv Myers, Zhu and Leskovec 2012 [12] Ñ ÷Ü Ë⌦oÇU(>§≤Ô-≥≠⇥⇡⇧Â\- Ü ↵⌦oÙc!ã Â¿flg Ë⌦oå. Ë⌦oÑqˇ⇥É|˛Í. 29⇧ ÜÍ. Ëãˆ. ≤Ô. Ñ‡. 71⇧ Ñ⌦oÔÂx‡º≤ÔÙc. ⇥'⇢xv. ÷Üi↵∫(≤Ô-#•ÑÙ. •qˇ⇥ v Shuai et al. 2012 [14] (_á!ãÜ0ó≤ÔÙ•#PÑ(6KìÑì•qˇõ⇥. 3. v÷. i↵í.

(12) (ÓMv>§≤ÔÑá{(≤Ô-Ñ>§qˇõ Ç⇢↵‘Ñ¶x Degree Betweenness Centrality. ~⇧v,œÜ≤Ô-Ñ↵‘. '⇢xv˝°(⌅.Â↵‘-√∫;Ñπ’ ↵‘Ñ¯—' Closeness Centrality -ì' Markov Centrality. ¨Ô+-√'. vqˇõBÑPÀl' [10, 17, 19]⇥Ñ. Ñ‹¬´↵\/>§≤Ô-. ↵‘(œ \⇧. ⇡.˙º b Graph-based Ñπ KìÑ‹¬Ñ˝õ (6⌥ H. ‹uÑ⌦o⇥d. bó’Ñπ’. Âœ. v÷v⌫f⌥(6KìÑq. ‹¬ evqˇõí ó’ [6, 9, 18]⇥6 ’:O·(6 User å H Project (⇡õ˙º. Entity. 1º. „Â⌥©«⌦. bó’Ñ˙. e˙!N↵. P6. ãÇ⇧ÓX2. ´-Ñ↵✏º Source Codes Ñq «ô –§Âå Commit Logs õ©«⌦(⌃êü>§≤ÔB Ô˝/à ˘<Ñ⌦o⇥. 2.2. ®¶!ãó’. 立. ‧ 國. 政治大 >§gËÑ±+qˇõ. )(œ. óÑ>§q. 學. d÷áÑ;ÅÓÑ/!Ï&œ. ⇡. ˇõÜÊ˛>§(6Ñ¢{¶Ñí ⇣,⇥ ⌘⌘ç∫®¶!ãÑ8√⌃ı⌥⌘⌘ Ñí ⇣,ÑÓÑ^8¯<⇥w‘Ü™ ®¶ó’/9⁄r(6ÑwÚ ⌅⌫f⌘(6®¶∞Ñ⇧Ó. F¡⇥ €Âq™. ⇡õ⌦. ÍÅ⌘⌘)((6q. Nat. ⌅. T\X2´Ñ‹. sit. ¬ ˙º®¶Ñó’!ã_ÔÂ~0(6Ñqˇõíè⇥. Ñ⇧Ó. y. ´ñ∫!Ï(6L∫ÑÕÅ⁄"⇥‡d. ‧. ⇧ÓÑú}. er. io. ®¶!ãÑÄS—~tÜÚì|Uó^8nM &´…(º⌅.>§Ñ… (⇥®¶˚q⇢NT\ ˙ºgπÑN˛ Â⇢.π✏ ⇣&( å8˙ ↵®. al. n. iv n C ¶Ñí ⇧Æ⇥⌘⌘ç∫®¶!ãå(6qˇõÑ0óKì W^<Ñ0π⇥ã hengchi U Ç ⌘⌘✏N(6Ñq «ôå⇧ÓÑgπÜ®¶∞Ñ⇧Óf(6 qˇõ0 óÑπ✏¶ Ñqˇ<⇥. 2.2.1 T. T. ✏N(6⌥⇧ÓNÄÑT\. N˛. Ü0ó(6. º∞Ñ⇧Ó[(. Collabrative Filtering. N˛Ñπ’/˙º6∆å⌃ê'œ. v÷^<(6ÑL∫. ⌅. ‹(6ÑwÚL∫⌥NªO}. &˙º. ⇣,(6•↵ÜÑL∫⌥⌦o⇥⇡.π’/G-∫⌘Nª. ÑL∫ O} ⌥⌥ÜÑL∫ O}¯ ⇥1⇢ó’⌥T N˛ÑÇı(º, œ®¶!ã-(6 ⇧ÓKìÑ¯<↵¶ ãÇ⇢SVD Gp<Èc⌃„⇥ T. N˛Ñ*fi(º⌘⌘. ⌘⌘(6®¶LÚ. s⌘⌘. ÅÂS(6 Ü„Û⇥ÑÚ®. ⌅-ÑL∫⌘(6®¶LÚ⇥FÇú. ⇧ÓÑ,Ígπ⇥ãÇ Ûø. ⌘⌘_ÔÂÂ(6wÚ. ∞(6†e«ô∆ 4. Çú⌘. T. N˛1!’c.

(13) 8Â\ ‡∫∞(6Ñ ⌅NºÌ⌘⇥. 2.2.2. Content-based Filtering. ˙ºgπÑN˛. ˙ºgπÑN˛/˙º⇧ÓÑgπœå(6↵∫Ñq «ôÜN˛˙ ÅÑ⌦ o⇥˙ºgπÑN˛Ñ;ÅÓÑ/⌫f®¶⌥(6NªúaÑ⇧Ó¯<Ñ∞⇧ Ó⇥(⇡.π’-. ñH⌘⌘ór÷⇧ÓgπÑyµh:b✏. I€⇣(⌘œzì-Ñyµ⌘œ⇥6å. ãÇ⌥⌅↵⇧Ó. ⌘⌘û⇧ÓÑU⇢Üu˙. ↵(6Ñ↵. ∫«ô⇥✏N⇡õyµ⌘œÑ⌦o ˚q1ÔÂ®¶v÷¯<Ñ⇧Ó⇥. n. ‡∫⇡.π’/˙ºgπÑq «ô É:O(6 v÷∫NªìWÑ⇤ ‡d˚q ˝®%⇧ÓÑ®< ¿fi _!’zö⇧ÓgπÑ¡Í*£⇥⇤. œ⌦⇡õπ’Ñ*:fi yfi⇥. 治 N˛å˙ºgπÑN˛ó’Ñ 政 ÜT 大. ó’t. ˜. ‧ 國. 立. 學. 2.2.3. ˜. Hybrid Algorithm. ãó’. ‧. '⇢x —Ñv/⌥T N˛å˙ºgπÑN˛D Ñ˜ π’⇥T N ˛ÑÓÑ/N˛â ⇢∫å⇧ÓKìT\Ñ⌦o \!✏ [15]⇥_h⌃„ ÑALÄS⇥‡∫É. ≈ÔÂ!ÏT. y. [13] ó’Úì⇣∫®¶!ã-. N˛ÑT\⌦o. io. n (_h⌃„ó’Ú ⌅⇥F1ºx⁄ : ↵OL X. _h⌃„". Fí. Ch. i Un. v. engchi. ↵§íÈcÜh:*ÂÑx<⇥ãÇ. å _h⌃„!ãÔÂ0œ(6 A. Û⇥ Y. (6å@. (6 A }ÜÛ⇥. ⇧ÓKìÑx<⇥6. £Ñ↵¶x<⇥. ⌘⌘ÔÂ((6⌥⇧ÓÑ§íÈcÜh:>§≤ÔÑT\. ⌦o⇥⌘⌘⌥ÈcgÑx< A T\. ⌥!Ï-Ñ(. ↵(6 ⇧ÓÑÈcÜ⇠⌅(6⌥⇧ÓKìÑw '⇢xÑ(6í yö⇧ÓÑwÚ ⌅⇥∫ÜK ⇡. }Û⇥ Y ⇥§íÈcÔÂ!Ï@. 9⁄⇡↵F∂. ˝. er. 6↵∫Ñú}ue!ãÑx“N↵-⇥. al. sit. Nat. Factorization Machines; Rendle and Steffen 2012. ñ∫(6å⇧ÓKìÑ‹¬↵¶⇥ãÇ. ;„Ñ⇧Ó∫(6 B ÑX2´. G⌘⌘((6. G-(6. ⇧ÓÑÈc-Â;„Ñ. !xx<⇠⌅r!T\Ñ‹¬⇥€Âq™ ⇡✏sW(6 B ✏NrX2´qˇÜ (6 A ⇥⌘⌘ç∫(6⌥⇧ÓÑ§íÈc-!Ï0óÑ<s/≤Ô-[(Ñqˇ õ⇥. 5.

(14) ,. ‡. vπ’ 治. ® ¶ ˚ q ⌥ > ⇤ T政 \ ≤ Ô Ñ 大^‘. 3.1. 立. ˛(6⌥. ↵…1N˛wÚ. ⌅. ˙º⇢↵. HKì±+‹¬ÑN↵⇥T. /✏NZ. N˛ÑN↵-. Nat. aœœ. n. Ñó’. a ¢{⇧ l. (≤Ô-Ñ>§qˇõ⇥. HÑg. \≤Ô-Ñ(6⌥ HKìÑ‹ \≤Ô-Ñ ¢{⇧ H Ñ‹. H Ñ‹¬⇥•W⌘⌘✏N_hx“Ê\T. ni CHh‹¬-Ñ[(qˇ< U engchi. 6. |. HÑO}⇥. er. io. …0®¶˚q-Ñ (6. \‹¬. ≈ÂS(6. ⇢↵(6Ñx«ÜÍ’⇣,(6 v÷. (T N˛ÑÇı⌦ ,v⌥>§T ¬^‘⇣®¶˚q⇥w‘Ü™ ⌘⌘⌥>§T ¬. HKìÑT. ‧. π. N˛/. ,Ñö©. y. T. /®¶˚q„€°(ÑÄS⇥(. sit. ⌦. Collaborative filtering. ‧ 國. N˛. 學. T. v. åœ. N˛. ˙œM¢{⇧.

(15) 3.2. [(qˇõÈcÑI€N↵. Contributor 1. Project 1. Contributor 2. Project 2. Contributor 3. Project 3. Contributor 4. Project 4. Project 5. a. R. c2. c3. c1. 1. 1. 0. c2. 1. c3. p5. 1. 1. 0. c2. 1. 1. 0. 0. 0. 1. 2. 3. 1. 0. 0. 1. 1. c3. 0. 0. 1. 0. 0. 1. 0. 1. 1. c4. 0. 0. 1. 1. 0. -. io. y. [ ¢{⇧ - ¢{⇧ ] Èc⌥ [ ¢{⇧ -. n. al. -. sit. Nat. \≤Ô-. c. ‧. b [ -. 3.1: >§T. p4. c4. Ch. 3.1∫ d ‡ ¿ I € N ↵ Ñ : ✏. G-(. i Un. v. ↵>§T. \≤Ô-. e n g c h i )Ñ∆ ⌘⌘ö©¢{⇧( Contributors. 3.1 a @: H( Projects )Ñ∆ m = |P | „h. Ç. ∫ C = {c1 , c2 , . . . , cn } ∫ P = {p1 , p2 , . . . , pm }  n = |C| „h¢{⇧Ñ∫x. HÑ↵x⇥•Wö© R ∫. 3.1 c. @. (ÈcgÑC. rij. ↵ n ⇥ m ÑÈc. R Èc/(Üh:>§≤Ô- [ ¢{⇧ -. :. H ] Èc. er. c4. 立0 1. 治p p p 政 1 c 0 1大 0. 學. c1. ‧ 國. A. H ] Ñ‹¬. Ç. Â ci ∫ pj Ñ¢{⇧K Grij = 1 Â / Grij = 0 ⇥çö© ↵ A ∫ n ⇥ n Ñ0•Èc Ç 3.1 b @: (Üœ¢{⇧|dKìÑ \‹¬ ( ÈcgÑC. aij. Â ci ⌥ cj i∫˛ì. \N. ↵Â⌦Ñ. H. G aij = 1. Â. / G aij = 0⇥1 Ü R ⌥ A i↵ÈcÑö©å f (·) ró✏ f (·) /(Ü Ç↵@:⇢. À. ,v–˙Ü. ↵[(qˇõÑI€˝✏. Å8e_h⌃„ó’ÑqˇõÈc2. f (·) ö©. 1. A Èc“⁄C Ñ<Ü-∫1 ÷6qˇõÈcgÑ<ÔÂ ⇢.Ñ- π✏(ãÇ⇢(6( ↵GitHub Fdπ’´Ù•…((,vq ÑF∂↵ ªU⌃v÷ ÑI€π✏⇥ 2. 7. H⌦Ñ–§!x).

(16) 8 > <1, f (rij ) = ?, > : 0,. if rij = 1, if 9 ck that i 6= k, aik = 1, and rkj = 1, otherwise.. (3.1). f (!). p1. p2. p3. p4. p5. p1. p2. p3. p4. p5. p1. p2. p3. p4. p5. c1. 0. 1. 0. 1. 1. c1. ?. 1. ?. 1. 1. c1. ?. 1. ?. 1. 1. c2. 1. 1. 0. 0. 0. c2. 1. 1. 0. ?. ?. c2. 1. 1. 0. ?. ?. c3. 0. 0. 1. 0. 0. c3. 0. 0. 1. ?. 0. c3. 0. 0. 1. ?. 0. c4. 0. 0. 1. 1. 0. c4. 0. ?. 1. 1. ?. c4. 0. ?. 1. 1. ?. a. b. c. 3.2: [(qˇõÈcÑI€N↵. 政治大. 立 ºœ. ‧ 國. „h. 3.2 a. Ç. ↵¢{⇧ ci. …. b. Û. ↵. @:. f (·) g. H pj Ñqˇ‹¬. .. ^ãÑ‹¬. 學. f (·) ÑI€N↵. ^ã⌃%∫⇢Ù•qˇ. )(Èc¯XÑó✏ T Èx˙ f (rij ) = 0. Nat. y. (8ef FM ˙!KM. Èx˙. sit. ‧. [(qˇ :qˇ⇥ R Èc(ó✏ (3.1)Ñ$I€å ⌘⌘ Ü ÀÑqˇõÈ c ç⌥Èc8e⌃„_hÑ!ã Ü0ó˙œ ↵[(qˇÑ<⇥. x“˙ÜÑ. ⇣H. T ÇÂ↵@:⇢. n. al. Ü7. er. f (rij ) = 0 /Åv\⌃„_h!ãÑ†bÑ◆ÙÊã. io. ÑC. Ch. engchi. i Un. v. (3.2). T = (tij ) = A · R 2 Rn⇥m . †b◆ÙÊãÑÈxN↵ Ç. 3.2 b. Û c. @: )(ó✏ (3.3)⇢. g(f (rij )) = g(rij0 ) = {0, if tij = 0, and rij0 = 0}. í. (3.3). ‰ tij = 0 ∫ T Pú-ÑC. G‰ f (rij ) = 0 &ç∫¢{⇧ ci (. ˚Uqˇõ⇥ÂPú tij > 0. Gç∫¢{⇧ ci (. ì•Ñqˇõ. s„h¢{⇧ ci ˛ì(. ˛ì \NÑ%4. (. H pj ⌦. H pj ⌦ @¢{⇥. 8. H pj ⌦. @¢{. H pj ⌦. Ô˝. Ù•. /¢{⇧ ci Nª.

(17) 3.3. ↵∫>§qˇõÑœ FM. ⌃„_h. /. ˝✏. .⇢(Ñ⇣,!ã. ÉP. Ü⌃„!ã(⇣,Pú⌦Ñÿ. ñ∫¶ Â yµÂ↵ÑH;'⇥(,v-Ñπ’- ⌘⌘(Ü¢{⇧⌥ H|dKì \Ñ‹¬ &⌥ HgM Ñ©⌦ot 0,vπ’ÑF∂ -. Â FM œ. œM¢{⇧Ñ>§qˇõ⇥. p1. p2. p3. p4. p5. c1. ?. 1. ?. 1. 1. c2. 1. 1. 0. ?. ?. c3. 0. 0. 1. ?. 0. c4. 0. ?. 1. 1. ?. p2. p3. p4. p5. Sum. c1. 0.5. 1.0. 0.6. 1.0. 1.0. 4.1. c2. 1.0. 1.0. 0.0. 0.4. 0.4. 2.8. c3. 0.0. 0.0. 1.0. 0.2. 0.0. 1.2. c4. 0.0. 0.6. 1.0. 1.0. 0.3. 2.9. 政治大. ‧ 國. 立. 學. a. p1. b. ‧. 3.3: Â⌃„_hœ. ↵∫>§qˇõ⌃xÑN↵. Contributor. y1. 0. x1. 1. 0. al. y2. 1. x2. 1. 0. y3. 0. x3. 1. y4. 1. x4. y5. 1. y6 y7. API information associated with contributor. Project. er. io. sit. y. Nat. target score. c. API information associated with project. 0.1. 0.3. 0.4. 0.3. 0.6. 0.1. 0.8. 0.2. 0.6. 0. 0. 0. 0. 0. 1. 0. 0. 0.2. 0.6. 0.1. 0.5. 0.1. 0.9. 1. 0. 0. 0. 0. 0. 0. 1. 0. 0.2. 0.6. 0.1. 0.7. 0.2. 0.7. x5. 1. 0. 0. 0. 0. 0. 0. 0. 1. 0.2. 0.6. 0.1. 0.9. 0.3. 0.1. 1. x6. 0. 1. 0. 0. 1. 0. 0. 0. 0. 0.4. 0.2. 0.5. 0.3. 0.4. 0.3. 1. x7. 0. 1. 0. 0. 0. 1. 0. 0. 0. 0.4. 0.2. 0.5. 0.8. 0.2. 0.6. c1. c2. c3. c4. p1. p2. p3. p4. p5. n. 0.6. 0. iv 0 1 0 0 0 0n 0.2 C h0e n1 g c0 h i0 U0 0.2 0. 0. sci. spj. 3.4: ⌃„_hÑ8e<✏ƒã. ↵∫>§qˇõ⌃xÑ Kå. Ç. 3.3 a. Û. b. 3.3@:. óN↵Ç. (⌃„_h0óå[(qˇõ. ⌘⌘⌥œ↵¢{⇧Ñqˇõ⌃xö©∫⇢œ↵. 9. H.

(18) pj fà¢{⇧ ci Ñ[(qˇõÑ=å. Ç. 3.3 b. c. Û. óÂ. ó. åœ. M¢{⇧Ñqˇõ⌃x⇥ Â,v- GitHub Ñ«ô∆∫ã. ↵⌫œÜÇU⌥«ô∆-Ñ⌦oI€∫. FM Ñyµ⌘œ⇥€↵&_ö©Ç↵⇢ • C: ¢{⇧Ñ∆ • P:. , ci 2 C, where i = 1, 2, 3, · · · , `. HÑ∆ , pj 2 P , where j = 1, 2, 3, · · · , m. • sci : ¢{⇧ ci Ñ API ©⌦o • sp j :. H pj Ñ API ©⌦o. G-(. ↵>§T. \≤Ô-. v-. ¢{⇧Ñ∆. C⌥. 政治大. HÑ∆. P. ⌘⌘⌃%⌥œ↵↵‘( Entity ) b Ñ©⌦oÑ&_ö©∫ sb v-↵‘( Entity )„ h M¢{⇧ / ↵ H⇥‘Ç sci „h¢{⇧ ci Ñ©⌦o spj „h H. 立HÑ©⌦oÑ≠¶. pj Ñ©⌦o. ¢{⇧⌥. ⌘⌘⌥ÊW-œ↵ÊãÂiˆ (ci , pj , sci , spj , f (rij )) Üö©. ‧ 國. 學. Ü⌦ö©Ñ&_. ⌃%1 dc ⌥ dp Üh:⇥. v- f (rij ) s/¢{⇧ ci ( H pj ⌦Ñqˇõ v<1˝✏ (3.1) ó ó⇥✏N †eyµÑä€˝✏ g(·) Üœû>§≤Ô†⌦©⌦o3 0 FM Ñyµ⌘œ. ‧. io. v- x(ci ,pj ) /iˆ (ci , pj ) ´ä€Ñyµ⌘œ. al. n. ⇡äxÜh:Êã-;çÑ¢{⇧⌥. Ch. Èc X -Ñ `-th ⌫. 1º n ⌥ m (å26Ñ⌥. H @Â u = n + m + dc + dp ⇥4. i Un. v. e n g c√ã˝✏ (3.4) hi. )(û¿,0ÑÊãu˙Ñyµ⌘œ [(Ñ>§qˇõ⇥G-⇣,OLÑ«ô Y 2 Ro œ. (3.4). sit. Nat. x(ci ,pj ) = g(ci , pj , sci , spj ) 2 Ru ,. y. g(·) Ç↵@:⇢. er. Ñä€. ( FM Ü!Ï 1yµÈc X 2 Ro⇥u å Ó⇡⌘œ (c ,pj ). œÜ¿,0Ñ◆ÙÊã x` i. ÊÍyµå y` = f (rij ) ∫ FM Ñ⇣,Ó⇡. FM 8eÑ<✏Ç. 3.4@:. _h- order-2 factorization machine Ñö©Ç↵@: [13]⇢ u u X u X X yˆ(x) := w0 + w i xi + hvi , vj ixi xj , i=1. !ã◆ÙÑN↵-. Å0. /i↵≠¶'✏ k Ñ⌘œÑgM( dot product )⇢ k X f =1. 3. ™. ⇥⌃„. (3.5). i=1 j=i+1. Ñ√x∫ w0 2 R, w 2 Ru å V 2 Ru⇥k. hvi , vj i :=. v- u ↵. vi,f · vj,f ,. ,v(Ñ GitHub «ô∆g †e0!ã-Ñ↵✏º©⌦o ⌥(‡¿4.1.1 ⇥ 4 ºœD¿,0ÑÊã =/ ↵;çÑ¢{⇧ … ↵;çÑ H⇥. 10. h·, ·i (3.6) Ùs0Ñ.

(19) ⇡·Ñ k 2 N+ 0 /ö©_h⌃„N↵-Ñ≠¶ÑÖ√x( hyperparameter )⇥ (˝✏ (3.5)-Ñ0 ↵⇢. ˝✏ yˆ(x) S(ci ) =. ⌘⌘ö©œ↵¢{⇧ ci Ñqˇõ⌃xÇ m X. (3.7). yˆ(x(ci ,pj ) ),. j=1. 立. 政治大. 學 ‧. ‧ 國 io. sit. y. Nat. n. al. er. ∫. (⌃„_h0óå[(qˇõKå ⌘⌘⌥œ↵¢{⇧Ñqˇõ⌃xö© œ↵ H pj fà¢{⇧ ci Ñ[(qˇõÑ=å⇥. Ch. engchi. 11. i Un. v.

(20) ,€‡ ÊWPú⌥ 4.1. «ô∆. ÷. 政治大. 立. Ñ Python ã|∫·. M30. «ôgπ. Â. Æ2. 30. (6Ñ. Æ⌥(6( Github ⌦. ‧. ‧ 國. 學. ∫Ü2L,vKÊW ⌘⌘ûT \s GitHub ⌦ê∆ÜÊW«ô⇥Â Python ↵✏û ∫; ⌘⌘Hù⁄ Git Awards ≤Ÿ1 Ñq - x˙Ü(h Lí h 4.1@:⇥«ôÑê∆π✏HÂ⇡ 30 ↵∫¡. ⌅Ñq. ÑlqX2´. 3,895 ↵(6⌥ 21,631 ↵X2´ H⇥3. n. al. dÜ(6⌥. HKìÑ. ⌥M!(Ñ‹⌦r˛x“ÑPú. 4.1.1. Bq. iv n C h e n g cÑ⌥«ô∆gÑ \‹¬K hi U. Modules ⌥Wˆ Packages †e˙!N↵Ñ!D⌥WˆÑ.^=xNºê' ⌅ 254,265 . ˆ†e Âq. ⇥. sit. Project. H. «ô∆g. =xq. er. \∫⌘⌘ÊW-«ô∆Ñ. io. X2´. y. Nat. Public Repository ∫˙ ~˙⇡ 30 ↵∫@ ˛ì \NÑ Python ã|⇧ \∫t‘«ô∆gÑ(6 User •Wçê∆⇡õ˛ì \NÑã|⇧Ñlq. H@(0Ñ!D. \∫©«⌦⇥1º´( ∫ÜMN˙!N↵Ñ⌥‹¶. @Â˙!ÑN↵&!⌥hË.^Ñ!D⌥W. (!xÈx˙´⇤⇢(6@. ( ⇤w„h'Ñ!D⌥Wˆ⇥. ↵✏º©⌦o. Èx!D⌥WˆÑ⌃è- ñHq Ü«ô∆g@ !D⌥Wˆ↵%Ñ ( Import ;á ó˙œ↵.^Ñ!D⌥Wˆ(«ô∆g´ (ÑsG!x⌅∫ 157.38 !. ÂsG. (!x\∫ù⁄. ⌥ÈxÑÄª<-∫150!. Í⌥. (;á. 1. http://git-awards.com/ ,áÍx«Â↵∫;äÑ(⇧ m ⇢∫DTÑ36 (⌘⌘Ñ«ô∆g⇥ 3 h⇤M 30 Ñ Python ã|∫·í Â =x∫ 3,895 ↵¢{⇧⌥ 21,631 ↵X2´ π ºÂ 2016 t 10 21 ÂÂMê∆å⇣⇥ 2. 12. HÑg.

(21) n. Ch. engchi. h 4.1: Git Awards ≤Ÿgí M30. !x'ºÄª<Ñ!D⌥Wˆ .Ñ!D⌥Wˆ. ⌘Û 1,075 .. y. sit. io. al. er. ‧ 國. 立. 政治大. Commit –§ =!x 11604 871 900 4098 3263 1013 128 463 1435 2900 2957 186 97 1645 5602 812 912 2844 924 152 422 107 4064 1140 2069 190 825 6347 5163 779. ‧. =x 48976 28822 28738 26277 25739 23164 20112 20065 13157 12947 12860 10294 10227 10036 9577 8673 8601 8594 8252 7480 7422 7260 7162 7099 6371 6147 6127 6026 4124 1789. ∞ÎNÑ H=x 71 9 20 13 32 7 11 21 4 5 21 2 5 6 31 6 9 18 43 4 7 5 27 74 24 37 13 30 80 52. 學. í 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30. Nat. (6 1 kennethreitz vinta jkbrzt donnemartin nvbn rg3 minimaxir josephmisiti Valloric fchollet chrissimpkins drduh faif soimort tomchristie binux apenwarr jonathanslenders miguelgrinberg alexjc fxsjy p-e-w coleifer mitsuhiko nvie DanMcInerney eliangcs stephenmcd alex toastdriven. ¡ Ñ H=x 49 9 13 10 32 7 11 21 4 4 21 2 5 5 18 6 9 16 39 4 7 5 23 61 18 37 11 27 41 38. i Un. v. Ñ Python ã|∫· Æ⌥q. \NÑ =∫x 772 217 878 519 162 406 41 196 119 271 20 26 57 79 625 36 41 93 119 3 31 5 175 579 217 37 66 574 786 457. «ô. e˙!Ñ©«⌦⇥ÈxÑPú⌥ü, 254,265 &Âd 1,075 .!D⌥WˆÑ. Æ. ö∫,v. Ñ↵✏º©«⌦ çÂ⌘œÑb✏†e˙!ÑN↵⇥ ÅË✏Ñ/. ˙!ÑN↵-Ñ©«⌦. (6⌥. H⇤2L↵%Ñq. ⇥(. 6©⌦o-Ñyµ< ∫r(6 \NÑ HX2´- (0Ñ&∫ 1,075 . ÆgÑ!D⌥WˆÑ (!xcè ⇥ HÑ©⌦o-Ñyµ< ∫ r. HX2´. (0Ñ&∫ 1,075 .. ⇥. 13. ÆgÑ!D⌥WˆÑ. (!xcè.

(22) 4.2. ÊW-ö. (ÊW-. Â Git Awards ≤ Ÿ ⌦ Ñ í. h 4.1@. \∫ÊWÑU0⇡ñ⇥Ç. : Git Awards Ñí Â(6¡ ÑX2´róÑ =xå\∫í ù⁄⇥, áÊWÑPú- _Ê (Üî. Ñ˙ñ Baseline ⌥,áπ’ÑPú Z‘⇤. M ↵˙ñ⌃%/⇢. 1. (6¡. Owned projects. Ñ H. 2. (6∞ÎNÑ H 3. (6(@. Ñxœ=å⇥. Written projects Commit. HÑ–§. Ñxœå⇥ !xÑ=å⇥. 治政大1998 [1] Ê i↵˙ñ/˙º PageRank Brin and Page 立 PageRank ó’gÑ¿fi ¿fiKìÑä⌃%Â↵πö©-ö ⇢. n. al. 1i⇧. ↵. \Ñ. HÑxœzö. ˛. er. io. H Gi↵(6KìÑäÑ⌦Õ< \NG∫0⇥. \. y. ä⌦⌦Õ<Ñ-ö⇢Âi(6˛ì. sit. Nat. 5. PageRank ó’≤ag. 0/1 -ö⇢Âi(6˛ì ˛ \NG∫0⇥. ‧. 4. PageRank ó’≤ag ä⌦⌦ÕÂå26 \ ↵ H Gi⇧ ä ä⌦⌦Õ<∫1. ⌥(6-∫. 學. ‧ 國. 4. Ch. i Un. v. ,áÑπ’ Â_h⌃„ÑWˆK ⇢ libFM Rendle 2012 [13] 2LÊ W v-√xdÜÌ„!x-∫ 200 v÷√xÜ∫ libFM ⇣-Ñÿç<⇥dd. engchi. 1º PageRank ⌥ FM ó’Ñ®_'. K. ,áÑÊWx⁄Ü∫ 20 !¯. √. xÑÊWPúÑsG<⇥. 4.3. U0⌥⇡. ÊWPúÑU0(Üi.U0í Ñ¯‹¬x˝✏ Spearman’s Rho (⇢) [11] å Kendall’s Tau (⌧ ) [5, 13] ÜU0ÊWPúÑ*£ ⌦i.ÑU0π’í. h˛. }⌥. {x1 , x2 , . . . , xn }. fi Ñ < À º 1 Û -1 K ì ⇥ Â - c ∫ Ñ í ⇣,Ñí. ↵⌫i↵˝✏⌃% ó⇢ 4. Pú∫ Y = {y1 , y2 , . . . , yn }. ,á( PageRank ÑÊW- damping factor Ñ<-ö∫ 0.85⇥. 14. Gí. Pú∫ X = PúÑ*£ù.

(23) P. (xi yi )2 , n(n2 1) #concordant pairs #discordant pairs ⌧ = 0.5 · n · (n 1) 6. ⇢ = 1. ( Kendall’s Tau ,œ$∑ùˆÇ↵⇢. ˚i↵í. PúÑM. (xi , yi ) å (xj , yj ) @lÑ∆. #concordant pairs ⇢Âˇ≥ xi > xj å yi > yj. x i < xj å y i < y j ⇥. #discordant pairs ⇢Âˇ≥ xi > xj å yi < yj. x i < xj å y i > y j ⇥. 立. ⇢ ⌧. Top 30. ⇢ ⌧. 0.8182 0.6444. 0.7818 0.6000. 0.3805 0.2842. 0.4045 0.3053. 0.4601 0.1684. 0.0029 0.0207. 0.1604 0.1172. al. n. Top 20. 0.7697 0.6000. io. ⇢ ⌧. Nat. Top 10. ¡ Ñ ∞ÎNÑ FM FM H=x H=x PageRank !API⌦o DAPI⌦o. Ch. 0.0207 0.0046. engchi. 0.8958 0.7533. 0.9533 0.8711. 0.5252 0.4458. 0.3938 0.3337. 0.0927 0.1195. 0.0242 0.0575. y. U0⌥⇡. sit. ÊWπ’. ‧. ‧ 國. ÊWPú. 學. 4.4. 政治大. / yi = yj ⇥. er. i ⇧ Ü ^ ⇢Â xi = xj. i Un. v. h 4.2: M30 Ñ⇣,PúU0⌃x ÊWπ’. U0⌥⇡. FM !API⌦o sG< äpx. FM DAPI⌦o sG< äpx. Top 10. ⇢ ⌧. 0.8958 0.7533. 0.001954 0.007584. 0.9533 0.8711. 0.000283 0.002474. Top 20. ⇢ ⌧. 0.5252 0.4458. 0.001070 0.000703. 0.3938 0.3337. 0.002217 0.001692. Top 30. ⇢ ⌧. 0.0927 0.1195. 0.002697 0.000664. 0.0242 0.0575. 0.000573 0.000316. h 4.3: 20 !ÊWU0⌃xÑsG<⌥äpx (ÊWh 4.2o:Ü. ⌘⌘⌥⇣,. óÑí. ⌥ Git Awards ≤Ÿ⌦Ñí. Ñ˙ñπ’⌥,áÑπ’ 15. 2L‘⇤⇥. ⌃%( Spearman’s rho (⇢) å Kendall’s.

(24) tau (⌧ ) i.U0˝✏Ñ⌃x⇥(. Ñ‘⇤˙ñ-. ÂœM¢{⇧∞ÎÑ. H=. xÜí ÑPú (íèM10 ÑPúó0 }Ñh˛⇥ Ç. h 4.2@:. MÑ˛aK. ,vÑπ’dÜ(†e↵✏º©«⌦ÑÑí. v⇠Ñ⇣,PúÜ(@. Ñü‡⌥⇤(‡¿ 4.5.12. e. FM !ãT0Ü }Ñh˛. 5. ‘⇤˙ñÑ⌃xK⌦. ¡Í. ↵. í. ¡Í↵M. ÷⌥⌃ê⇥(†e©«⌦ÑPú-. Top 10 Ñ. s ⇢ = 0.9533 ⌥ ⌧ = 0.8711⇥. 1º FM !ãó’Ñ®_' ÊWÑPúö∫ 20 !¯ √xÊWÑsG < Ç 4.3@: ÊWPúÑäpx Variance <à✏ F(ƒ 0.0002 ⌥ 0.002 Kì „hÊWPú. 6ÇÛiö⇥. (⇧. H 0.141090137 0.133651760 0.127058477 0.121183140 0.119622888 0.116999270 0.115921046 0.110107542 0.110074160 0.109540859. y. Nat. sit. io. n. al. Python modules/packages Ñ†⌦⌦Õ. er. h 4.4: M10 Ù2 ⌦ÕPú. ‧. ‧ 國. 立. 政治大. 0.167783150 sqlalchemy.types† 0.140377013 pygments.highlight§ 0.136406151 decimal.Decimal 0.133800450 asyncio 0.128256130 jinja2.runtime.Undefined§ 0.127342390 django.views.generic.View§ 0.122519923 django.test.TestCase§ 0.120499280 sqlalchemy.ForeignKey† 0.112403762 module base.ModuleBase 0.111670590 django.template.loader§. 學. numpy⇤ twisted.cred.portal§ flask.current app§ django.core.paginator.Paginator§ django.contrib.sitemaps.Sitemap§ scipy.optimize⇤ weboob.tools.backend.Module§ curses constants.eStart core. v. e¿fl Python modules/packages Ñ©«⌦( FM !ã◆ÙB†⌦Ñ h 4.4-ÑPú ÂOrÜ\∫ (⌦Ñ⌃^. 1. —x ó!D⇤. Ch. engchi. i Un. Scientific computing modules. 2. «ô´…(¯‹!D†. Database-related modules. 3. ≤ ⌥≤ŸÑ…(ã|!D§. Web-related modules. v-⇧r⇡⇠Ñ!D⇢´(º—x. ó. Xr⇡⇠Ñ!D⇢´(º«ô. Database Õ\¯‹ Õr⇡⇠Ñ!D⇢´(º≤ ⌥≤ŸÑ…(ã h 4.4Ñí PúÔÂ¿fl˙ ≤ ã|¯‹Ñ!D/h-Ñ;Å!D 1. ´ |. dÔÂÓM⇥ Ñ Python ã|∫·@‹ËÑã|Ô˝⇢∫≤Ÿ⌥≤ …(⇥ 5. ÊWÑPú⌥‘⇤˙ñÑoW'¢W. Significance Test. 16. Pú P <Ü✏º 0.05.

(25) ñ∫. 立. ‧ 國. 學 y. ‧ er. ⌥«ô⌥ÊWPúÂ¿fi≤ÔÑb✏Zñ∫H˛. al. n. ≤Ÿüã. Python ã|∫·Ñí. ⌘⌘–õÜ«ô∆-M10. io. 6. 4.1: ñ∫ ≤Ÿ∂À. Nat. (,‡¿-. 政治大. sit. 4.5. ˙⌥‡¿ 4.4¯|…ÑPú Ñ(π✏⌃∫. ↵@J⇥. iv ñ∫ Â≤ Ñb✏\H˛ n C hengchi U. Ç. ⇣,PúÑñ∫ &(¿flKåx 4.1@:. ≤. 1. (6⌥ HÑ‹¬≤Ô 2. s0«⌦ 3.. ⇣,π’Ñ⌥€. ((6⌥ HÑ‹¬≤Ô- (6Â√rÑ¿fih: HÂÕrÑ¿fih : qˇõ‹¬Â¿fiKìÑ#⁄h: ⇣,ÑPú (6ÂróÑ⌃xÿ N. ⌃è1ÊÛÛZ*Híè⇥(s0«⌦Ño:@J-. (6. H. 1. c∫Ñí. Pú⌥ÊWπ’Ñ⇣,Pú⇥,v(ñ∫. –õÜ . Ñπ’ÜZ⇣,PúÑ‘⇤ FM †⌦↵✏ºÑ©yµ⇥ 6. o:ÓM⌥⇡x«Ñ. d∫≤ŸüãÑ#P⇢https://clip.csie.org/github/. 17. Ñ≤. .π’⌃%/ Origin FM ⌥.

(26) Ç. 4.2@:. ( Origin π’-. o:ÑPú/Â∞ÎNÑ. H\∫‘⇤˙. ñ í Ñ⇣,⌃x/q (6∞ÎNÑ H=x (≤Ô-ÔÂ¿fló0Ñ/ ¿fiKìÑ‹¬#⁄xœ⌥∆¶ ‡∫íèÑπ’@Ù ⇧Z01ÊÛÛ^ Ñ ˛a. _⌥ùd˛aÜ‘⇤v÷i↵π’Ññ∫Hú⇥ (‘⇤. 4.2⌥. 4.3Ññ∫Pú-. ÷6π’ FM ⇣,Ñí. Pú⇤s. F. ñ∫⌦ÔÂ¿fl˙#PÑ∆¶Â⇤G˚ ‹ÇÑπ✏H˛⇥ ÕÅÑ/(‘⇤ 4.3⌥ 4.4Ññ∫Pú- ÔÂ¿fl˙( M «⌦ÑÊW-ró }Ñí #PÑ⌃H≈b_⌥. 立. 4.2Y∫¯— Â1ÊÛÛ^ Ññ∫Hú⇥. 政治大. 學 ‧. ‧ 國 io. sit. y. Nat. n. al. er. Pú. Ch. engchi. i Un. v. 4.2: Â∞ÎNÑ H=xíèÑ⇣,Pú. 18.

(27) 4.3: FM !†eM «⌦Ñ⇣,Pú. 學 ‧. ‧ 國. 立. 政治大. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. 4.4: FM †eM «⌦Ñ⇣,Pú. 19.

(28) 4.5.1. í. (¿fl. ¡Í↵MÑ˛a⌃ê. 4.3⌥. 'E¶Ñ=Ó a. 4.4Ñ‘⇤(⇤nM. F/(†eM. ÔÂ|˛,å. «⌦KM. «⌦ÑÊW-. r r. vinta. Ñ(6. ⇣,Pú. W. (6(‹¬≤Ô-&!y%Ñ=Ó˛ (6ÑÊWPú{". Ü'E¶Ñ. e. ≈b ∫Ü≠„" eÑü‡ @ÂÙ2 eÑ¿fld (6(«ô∆-Ñq x⁄ q PúÇ h 4.5@: ÔÂ|˛v- awesome-python (652070) H Ñ. x⌥!DWˆÑ.^xKìÑ‹¬. =. x 28,822 -. %ºv÷Ñ. (6 vinta ¡. H. H awesome-python (652070) 1Â. Ñ. x 28,640 ThËÑ. 99.4% ( †eM «⌦ÑÊW- awesome-python (652070) HÕÔ´ñ∫ ↵c8Ñ H F/(†eM «⌦ÑÊW- ‡∫ awesome-python (652070) (x∫ 0. HÑ!D⌥Wˆ. M. «⌦Ñ⌦Õó(6. å†=Ñ⇣,⌃x'E. e≈b2 qˇt‘ÑM 10 ⇣,⇥(d 治政 ¿flKå⌥¡ ÿ =x F/!˚U (!D⌥Wˆ 大 / (=xu⌘Ñ≈ b Ç H awesome-python(652070) 立 ñ∫ ºã|…(Ñ Python H ⌥d ↵Ó. @ÂÊWPú". 'E¶Ñ. M10. Ñ¿flPú. ¿flÑPú|˛M 20. ⌥t⌃«⌦Yt^Ñ. -¡. «⌦Yt^Ñ. H. Â. HÑ(6. ‧. (6. ‧ 國. 1ñ∫. 學. HÑ'Íx^º«⌦YtÑ^ã⇥. #Ñπ✏¿fl 2M. drduh ⌥. y. Nat. faif #(†eM «⌦Ñ!ã- " Ü⇣,⌃x'E eÑ≈¡ @ÂM⇤ ‡¿ 4.4g h 4.2ÑÊWPú-M 20 Ñ⇣,U0 (†eM «⌦å". ↵∫–§ H 1 =!x BlackWidow(652070) 64 Haul(652070) 54 anobii2douban(652070) 4 awesome-python(652070) 599 django-email-confirm-la(652070) 85 pangu.py(652070) 36 python-pay2go(652070) 8 python-smsking(652070) 3 sbi.py(652070) 18. n. al. Ch. H–§ =!x 65 55 4 980 101 39 8 3 19. er. io. sit. Üí ¡Í↵MÑ≈b⇥. n U engchi. ã|√⌥ v i =∫x 2 2 1 571 4 55 1 1 2. h 4.5: (6 vinta ¡ Ñ HÑq «⌦. 20. !DWˆ =x .^x 5 13 80 15 3 6 28,640 0 11 64 4 4 3 13 2 6 23 8.

(29) ,î‡ P÷ P÷ ↵π’. )((6å⇧ÓÑ◆oÜ!Ï>§T. \≤a-Ñ. 學. ,v–˙Ü. 政治大. 立. ‧ 國. 5.1. (6(>§-Ñqˇõ⇥&Â Python û. Nat. -Ñ«ô∆2LÊW. ( GitHub ≤Ÿ. y. (qˇÜœ. ‧. qˇõÙc ✏N⌥>§T\≤a⌥®¶˚q^‘Üœ (6Ñqˇõ⇥w‘ Ü™ ⌘⌘⇢N,œ˙º⌃„_h FM ÑF∂↵ œ↵(6 ºœ↵⇧ÓÑ[. ✏N†e© M10MÑã|∫. ∫;. ÊWÑPúIÊÜ@–˙Ñπ’Ñ. sit. al. C dÜ⌃êÊWPúÑx⁄K h KåÑ!ã-ÑM. «⌦. eÂñ∫ n g c h≤i. º>§T\≤aÑ. iv n U ÑH˛Pú. Ñ. ˘<Ñ⇥. ÜÍ GitHub Ñ Java «ô º!D⌥Wˆ©·o. / H ÷@ÑáW⌦o ⇥. 21. &⌃ê†e◆Ù. flå⌃„/^8. *Üπ⌘⇧Ï†ev÷û .^Ñ«ô∆Êã ãÇ ∆ ⇥ddK (*ÜÑÊW-⌥⇧Ïå⌃êv÷ ãÇ –§ÑBì·o. ∫⇥. er. s ⇢ = 0.9533 ⌥ ⌧ = 0.8711⇥. n. ·. io. Python ã|∫·–õÜ‘~.˙ñπ’Ù}Ñí ⇥ddK ·o FM !ã2Ù eT⇣Ü 'E¶ÑH˝h˛ s(í. H'.

(30) √⇤á{ [1] S. Brin and L. Page. The anatomy of a large-scale hypertextual web search engine. In Proceedings of the 7th International World Wide Web Conference, pages 107–117, 1998.. 政治大 mation propagation in the flickr social network. In Proceedings of the 18th Interna立 tional Conference on World Wide Web, pages 721–730. ACM, 2009.. [2] M. Cha, A. Mislove, and K. P. Gummadi. A measurement-driven analysis of infor-. ‧ 國. 學. [3] D. Gruhl, R. Guha, D. Liben-Nowell, and A. Tomkins.. Information diffusion. through blogspace. In Proceedings of the 13th International Conference on World. ‧. Wide Web, pages 491–501. ACM, 2004.. y. Nat. [4] L. Hong, O. Dan, and B. D. Davison. Predicting popular messages in twitter. In. sit. Proceedings of the 20th International Conference Companion on World Wide Web,. n. al. er. io. pages 57–58. ACM, 2011.. i Un. v. [5] M. G. Kendall. A new measure of rank correlation. Biometrika, 30(1/2):81–93, 1938.. Ch. engchi. [6] N. Li and D. Gillet. Identifying influential scholars in academic social media platforms. In Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, pages 608–614. ACM, 2013. [7] A. Lima, L. Rossi, and M. Musolesi. Coding together at scale: Github as a collaborative social network. arXiv preprint arXiv:1407.2535, 2014. [8] L. Liu, J. Tang, J. Han, and S. Yang. Learning influence from heterogeneous social networks. Data Mining and Knowledge Discovery, 25(3):511–544, 2012. [9] X. Liu, J. Bollen, M. L. Nelson, and H. Van de Sompel. Co-authorship networks in the digital library research community. Information Processing & Management, 41(6):1462–1480, 2005.. 22.

(31) [10] P. Mutschke. Mining networks and central entities in digital libraries. a graph theoretic approach applied to co-author networks. In International Symposium on Intelligent Data Analysis, pages 155–166. Springer, 2003. [11] J. L. Myers, A. Well, and R. F. Lorch. Research design and statistical analysis. Routledge, 2010. [12] S. A. Myers, C. Zhu, and J. Leskovec. Information diffusion and external influence in networks. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 33–41. ACM, 2012. [13] S. Rendle. Factorization machines with libfm. ACM Transactions on Intelligent Systems and Technology (TIST), 3(3):57, 2012.. 政治大. [14] X. Shuai, Y. Ding, J. Busemeyer, S. Chen, Y. Sun, and J. Tang. Modeling indirect influence on twitter. International Journal on Semantic Web and Information Systems (IJSWIS), 8(4):20–36, 2012.. 立. ‧ 國. 學. [15] L. Terveen and W. Hill. Beyond recommender systems: Helping people help each other. HCI in the New Millennium, 1(2001):487–509, 2001.. ‧. [16] M.-F. Tsai, C.-J. Wang, and Z.-L. Lin. Social influencer analysis with factorization. sit. y. Nat. machines. In Proceedings of the ACM Web Science Conference, WebSci ’15, pages 50:1–50:2, New York, NY, USA, 2015. ACM.. n. al. er. io. [17] S. White and P. Smyth. Algorithms for estimating relative importance in networks.. v. In Proceedings of the ninth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 266–275. ACM, 2003.. Ch. engchi. i Un. [18] E. Yan and Y. Ding. Discovering author impact: A pagerank perspective. Information Processing & Management, 47(1):125–134, 2011. [19] L.-c. Yin, H. Kretschmer, R. A. Hanneman, and Z.-y. Liu. Connection and stratification in research collaboration: An analysis of the collnet network. Information Processing & Management, 42(6):1599–1613, 2006.. 23.

(32)