4.1 ໦ᆄीᆉ

໦ᆄၮᆉȐCloud Computingȑࡰޑࢂ΋ᅿஒीᆉᆶᓯӸ৒ໆբࣁ୍ܺǴ٠ࡪ ሡ؃ගٮ๏౦ᄬȐHeterogeneousȑޗ୔ޑಖᆄௗԏޣǶ໦ᆄीᆉڀԖΠӈਡЈ੝

܄[22]Ǻ

z ௵௘܄ׯ຾٬Ҕޣख़ཥϩଛ୷ᘵ೛ࡼޑૈΚǶ

z ᔈҔำԄϟय़ޑёၲ܄٬೬ᡏᆶ໦ᆄҬϕ൩ႽΓᐒϕ୏ϟय़΋ኬǴѬ೯த٬

Ҕ୷ܭ REST ࢎᄬޑ APIsǶ

z ԋҁफ़եࢂӢࣁӧϦԖ໦ޑ໺ᒡኳԄύЍ࡭ς࿶ᙯᡂࣁᔼၮԋҁǶ

z ೛ഢکՏ࿼ᐱҥᡣ٬Ҕޣёа٬ҔᘤំᏔӸڗس಍ǴԶόሡाӧЯԾي٬Ҕ Ֆᅿး࿼܈يೀՖӦǶ

z ຀ᔕϯמೌϢ೚՛ܺᏔکᓯӸး࿼Ӆ٦کճҔ౗ቚуǴԶᔈҔำԄࡐ৒ܰவ

΋ঁჴᡏ՛ܺᏔᎂ౽ډќ΋ঁǶ

z ӭચЊ٬౲ӭ٬ҔޣӅ٦ၗྍᆶԋҁǴவԶၲډ୷ᘵ೛ࡼӧࢌೀեԋҁ໣໣ύ ϯᆅ౛ȐٯӵǺ෧Ͽ܊ӦౢǵႝΚ฻ȑǴঢ়ঢ়ॶॄၩૈΚቚуȐ٬Ҕޣόሡाࣁ ёૈޑനଯॄၩ฻ભ຾Չࡌ೷ȑǴճճҔ౗ᆶਏ౗ቚуȐа߻س಍࿶தѝԖ 10%

ډ 20%٬Ҕ౗ȑǶ

z ёё᎞܄ቚуࢂӢࣁёҔ໦ᆄؼӳޑ೛ीࡌ࿼ӭঁഢҔઠᗺǴ፾ӝҾ཰҉ុ࿶

ᔼᆶؠᜤൺচǶ

z ёᘉ৖܄کቸ܄࿶җ୏ᄊȐࡪሡ؃ȑӧ΋ঁؼӳಈࡋȐFine-grainedȑ΢ޑၗ

ྍٮ๏Ǵௗ߈ܭջਔޑԾך୍ܺǴค໪٬Ҕޣჹঢ়ॶ຾Չࡌ೷Ƕ

z ܄ૈ೏ᅱ௓ǴЪ٬Ҕᆛၡ୍ܺ଺ࣁس಍ϟय़ࡌ೷΋ठ܄ᆶ݊ණጠӝࢎᄬǶ z Ӽӄ೏ׯ๓Ǵҗܭၗ਑ޑ໣ύϯǴቚуΑӼӄၗྍ฻฻Ǵՠჹ੝ۓ௵གၗ਑

Ѩѐ௓ڋǴаϷલЮӼӄޑᓯӸਡЈሡा೏࡭ុᜢݙǶ

໦ᆄ୍ܺёϩࣁΟᅿቫԛǴΨ൩ࢂ೬ᡏǵѳѠаϷ୷ᘵ೛ࡼǴӵΠǺ

z ೬ᡏջ୍ܺȐSoftware as a ServiceȑǺٮᔈ୘ӧ໦ᆄ΢Ӽး٠ᏹբᔈҔ೬ᡏǴ Զ٬Ҕޣவ՛ܺᆄӸڗ೬ᡏǴό໪ᆅ౛໦ᆄ୷ᘵ೛ࡼکᔈҔำԄၮՉѳѠǴ Ψό໪ӼးکၮՉ໦ᆄ΢ޑᔈҔำԄǴѝाೱ΢ᆛၡ൩ёа٬ҔǶٯӵǺ Google Apps ک Yahoo คӜλઠǶ

z ѳѠջ୍ܺȐPlatform as a ServiceȑǺٮᔈ୘ගٮ໦ᆄѳѠ๏٬ҔޣǴځх֖

բ཰س಍ǵำԄᇟق୺Չᕉნǵၗ਑৤ᆶ՛ܺᏔǴᔈҔำԄ໒วޣёаӧ໦

ᆄѳѠ΢໒วکၮՉдॺޑ೬ᡏǴό໪޸ԋҁӧᖼວᆶᆅ౛ፄᚇޑฯᡏک೬ ᡏቫǴԶԖ٤ѳѠϷ୍ܺ߾཮Ծ୏٩ྣᔈҔำԄሡ؃ϩଛۭቫޑीᆉᆶᓯӸ

ၗྍǴᡣ٬ҔޣόሡाЋ୏ᆅ౛ǶٯӵǺGoogle App Engine ک Microsoft Windows AzureǶ

z ୷ᘵ೛ࡼջ୍ܺȐInfrastructure as a ServiceȑǺന୷ҁޑ໦ᆄ୍ܺኳࠠǴٮᔈ

୘ගٮჴᡏႝတ܈຀ᔕᐒᏔǵᓯӸᡏǵٛОᕅǵॄၩѳᑽᏔکᆛၡǶ୷ᘵ೛

ࡼջ୍ܺٮᔈ୘٩ሡ؃வεࠠၗ਑ύЈύගٮ೭٤ၗྍǶ٬ҔޣӼးբ཰س

಍ࢀႽᔞȐImageȑаϷдॺޑᔈҔ೬ᡏӧᐒᏔ΢Ǵ٠ॄೢᆢঅǵߥᎦբ཰س

಍کᔈҔำԄǶٯӵǺAmazon EC2 ک IBM Blue CloudǶ

ҁፕЎ٬Ҕ ASP.NET MVC 4 Developer Preview ᆶ C# ໒วس಍Ǵ٠೽࿿ډ Windows Azure Platform ᆶ SQL AzureȐ୷ܭ໦ᆄीᆉǵёᘉ৖ހҁޑ SQL ՛ܺ

ᏔȑǶ

4.2 س س಍ࢬำ

௢ᙚس಍ӧ೭ঁس಍ύתᄽΑᜢᗖفՅǴॄೢᔠ઩٠ᓯӸޗҬკᆶ໒ܫკǴ ٠٩Ᏽঁձ܈იᡏ٬Ҕޣޑፎ؃௃ნǴౢғᇯЬᅰ៿ޑᓓ᡺௢ᙚ่݀ǹ௢ᙚس಍

ёϩࣁᔠ઩ᆶ௢ᙚٿঁኳಔǴس಍ࢬำӵკ 4.1Ƕ

ӧᔠ઩ኳಔύǴ२ӃǴঁձ܈იᡏ٬ҔޣЋ୏ᒡΕ܈س಍Ծ୏ว౜྽߻௃ნǴ ӛس಍ፎ؃ࡌ᝼Ǵس಍཮ڥћ Graph API ڗளҁԛፎ؃܌ሡ٬ҔޣکӕՔޑᐕўၗ

਑Ȑ٬Ҕޣᔞਢǵᓓ᡺ၗૻǵѺьၗ਑ᆶޗҬკȑǴவύှ݋کೀ౛؂ԛӣᔈޑ JSON

਱Ԅϐၗ਑ǴஒځකΕ܈׳ཥډၗ਑৤ύǴ٠ќ໒ୋጕำȐThreadingȑፎ؃܌Ԗ

ܻ϶ޑᐕўၗ਑аᙦ൤ၗ਑໣ǹ྽ၗ਑৤ڗளҁԛፎ؃܌ሡၗ਑ࡕȐୋጕำёૈ

ۘ҂่״ȑǴ཮Ӄ೏Ⴃೀ౛ԋёҥջ٬Ҕޑၗ਑໣Ǵ٠ڥћ௢ᙚኳಔǶ

ӧ௢ᙚኳಔύǴ཮Ӄ٬ҔՏ࿼ᆶຯᚆୖኧவၗ਑໣ύၸᘠሡԵቾޑᓓ᡺Ǵа ቚமس಍ਏૈǴ٠வၗ਑৤ύᕴ่ҁԛፎ؃ޑޗҬკᆶ௃ნၗૻǴฅࡕճҔക࿯

3.3.2 ܌ගډޑБำԄǴႣෳӧᕴ่ޑޗҬკᆶ௃ნၗૻΠǴ٬ҔޣᆶӕՔჹ؂ৎ ᓓ᡺ӅӕޑୃӳຑϩǴനࡕౢғ΋ঁ࿶җႣෳୃӳຑϩ௨ׇޑ n ৎᓓ᡺మൂǴ٬

ځૈ୼ᡉҢӧՉ୏း࿼ޑλᑻჿ΢Ǵගٮ྽Πന٫ޑࡌ᝼Ƕ

კ 4.1 س಍ࢬำკ

Call Graph API for user and all companions’ data

Web service returns data in JSON format

More friends’

data available Call Graph API for friends’ data

Web service returns data in JSON format

݌ݎ݁݀݅ܿݐ(ܥܵ, ݌, ݐ) =σ_௨א஼ௌσ_௨ᇲא௎ݏ݅݉(ݑ, ݑ^ᇱ, ݐ)× ݏݑ݉(ݑ^ᇱ, ݌, ݐ) Predict user and all companions' rating for each restaurant in the context info and social graph

Generate a list of restaurants sorted by predicted rating with size n Stop

Decode JSON and insert/update processed data into DB

Decode JSON and insert/update processed data into DB

DB

Use location and distance parameter to filter restaurant from DB

Stop Input:

The user’s request with context

Use the user’s request to summarize context and social graph from DB

Recommendation Module Preprocessing

Yes

No

Retrieval Module

4.3 ٬ ٬Ҕޣϟय़

ҁس಍ϐ٬Ҕޣϟय़ᆶ Google Map షམǴځගٮ JavaScript API ᡣ໒วޣ᏾

ӝӦკډ፾Ӧ܄୍ܺύǹךᗋ٬Ҕ JavaScript ᆶ jQuery բࣁ࠼Њᆄဌҁᇟقቚ຾

Γᐒϕ୏ϟय़Ǵ٠٬Ҕ AJAX מೌᆶ JSON ໺ᒡǴ٬س಍׳ࣁِ௘Ӧӣᔈ٬Ҕޣ ޑ୏բǶ

ӧϕ୏ԄޑӦკϟय़ύǴس಍཮೸ၸᘤំᏔ٬ҔޑۓՏמೌԾ୏ว౜٬Ҕޣ

྽߻ޑՏ࿼٠߇ӣ࣬ᜢၗૻǴ٬Ҕޣёᗺᔐߕ߈ᓓ᡺మൂύޑ໨Ҟ܈Ӧკ΢ޑ኱

૶࣮ࢗಒ࿯ǴΨёаܦԕӦკѐ൨פᓓ᡺ǹ܈ϪඤԿ௢ᙚኳԄǴԾ୏೛ۓ܈Ћ୏

অׯ྽߻ޑՏ࿼ᆶ௃ნǴӛس಍ፎ؃ࡌ᝼Ǵӵკ 4.2ǵკ 4.3 کკ 4.4Ǻ

კ 4.2 ٬Ҕ Web ϟय़ᘤំӦკک൨פᓓ᡺ޑᄒკ

კ 4.3 ٬Ҕ Web ϟय़ፎ؃ࡌ᝼ޑᄒკ

კ 4.4 ٬Ҕ Mobile ϟय़ޑᄒკ

5 ჴ ჴᡍᆶຑሽ

௢ᙚس಍നத٬ҔҬΰᡍ᛾Ȑcross-validationȑᡍ᛾௢ᙚᄽᆉݤޑ܄ૈǴٯӵǺ Sarwar ฻Γ[23]٬Ҕٿಔ੿ჴШࣚޑၗ਑໣ᡍ᛾൳ᅿ௢ᙚᄽᆉݤޑ܄ૈǴϩձࣁႝ

ቹ௢ᙚᆛઠᆶεࠠႝη୘୍ޑҬܰइᒵǹдᒿᐒܜڗ MovieLens ᆛઠຑϩ 20 ԛа

΢ޑၗ਑Ǵளډ 943 ঁ٬Ҕޣჹ 1682 ೽ႝቹޑ 10 ࿤ԛຑϩǴаϷ Fingerhut ႝη

୘୍ϦљޑᐕўҬܰइᒵǴх֖ 6502 Տ࠼Њᖼວ 23554 ᅿౢࠔޑ 97045 ԛᖼວइ ᒵǴ٠ᔈҔ 5 ७Ҭΰᡍ᛾ѐ՗ीس಍ޑᆒዴࡋǵєӣ౗ᆶ F1 ࡰ኱ǶฅԶჹܭ΋ঁ

Չ୏௢ᙚس಍ԶقǴόૈѝࢂൂપ౒ෳ٬Ҕޣࢂցୃӳ೭ঁ໨ҞǴԶࢂѸ໪٩Ᏽ

٬Ҕޣୃӳޑำࡋ௨ׇǴ٠ڗ߻൳ঁ໨Ҟբࣁ௢ᙚమൂǴ٬Ѭ፾ӝᡉҢӧՉ୏း

࿼ޑλᑻჿ΢Ƕ

ӧҁჴᡍύǴךஒᢀჸ٬Ҕޣӧ௢ᙚమൂύჴሞᒧ᏷ᓓ᡺ޑՏ࿼Ǵຑሽس಍

܄ૈǹ୷ҁ΢Ǵѳ֡Տ࿼ຫ΢य़ж߄܄ૈຫଯǶ

5.1 ၗ਑ԏ໣

ҁჴᡍύޑၗ਑җ 69 Ӝ Facebook ٬ҔޣගٮǴԏ໣୔ୱ໣ύӧεѠчӦ୔Ȑך ޑғࢲ୮ȑǴԏ໣ډ 2010/8/25 Կ 2012/4/30 ޑѺьၗ਑Ǵ߄ 5.1ǵ߄ 5.2ǵ߄ 5.3ǵ კ 5.1 کკ 5.2 ඔॊΑ೭ঁၗ਑໣ޑ੝ᗺǺ

໨

კ 5.1 ԏ໣ډޑ܌Ԗᓓ᡺ϐϩѲკ

໨

໨Ҟ ѳѳ֡ॶ ύύՏኧ ኱኱ྗৡ നനεॶ നനλॶ

ڙෳޣܻ϶ኧ 504.01 423 346.48 2200 34 n=69

ڙෳޣԃស 23.28 22 4.99 49 18 n=69

٬Ҕޣԃស 23.52 22 5.87 107 13 n=25184 මჹᓓ᡺Ѻьޑ٬Ҕ

ޣԃស

22.99 22 4.17 106 15 n=3928

٬Ҕޣჹᓓ᡺ޑѺь ԛኧ

0.34 0 0.99 12 0 n=25184

٬ҔޣѺьၸޑᓓ᡺

ঁኧ

0.33 0 0.93 10 0 n=25184 ᓓ᡺೏Ѻьޑԛኧ 3.23 2 6.31 224 1 n=2691 මჹᓓ᡺Ѻьޑ٬Ҕ

ޣჹᓓ᡺ޑѺьԛኧ

2.21 2 1.48 12 1 n=3928 මჹᓓ᡺Ѻьޑ٬Ҕ

ޣѺьၸޑᓓ᡺ঁኧ

2.10 2 1.35 10 1 n=3928 ᓓ᡺Ѻьޑᓓ᡺೏Ѻ

ьԛኧ

1.05 1 0.29 10 1 n=8264

ᓓ᡺ѺьޑѺьޣԃ ស

22.96 22 3.97 106 15 n=8264 ᓓ᡺ѺьޑӕՔѳ֡

ԃស

23.55 23 3.91 107 16 n=8264 ᓓ᡺ѺьޑӕՔ܄ձ

2010/8 2010/10 2010/12 2011/2 2011/4 2011/6 2011/8 2011/10 2011/12 2012/2 2012/4

ᓓ᡺Ѻьኧໆ

კ 5.2 ಍ीԏ໣ډޑᓓ᡺Ѻьၗ਑

Friend Classmate Colleague Family

ᓓ᡺Ѻьኧໆ

z ڙෳޣѳ֡Ԗ 504 ঁܻ϶Ǵ܌ԖΓޑѳ֡ԃសऊ 24 ྃǹ؂ঁΓѳ֡ჹᓓ᡺Ѻ ь 0.34 ԛǴѳ֡Ѻьၸ 0.33 ৎᓓ᡺Ǵ؂ৎᓓ᡺ѳ֡೏Ѻьऊ 3 ԛǴමჹᓓ᡺

ѺьޑΓѳ֡ჹᓓ᡺Ѻьऊ 2 ԛǴѳ֡Ѻьၸऊ 2 ৎᓓ᡺ǹᓓ᡺ѺьޑӕՔ ѳ֡ऊ 3 ΓǴѳ֡ԃសऊ 24 ྃǴζ܄К౗ัଯǴӵ߄ 5.3Ƕ

z ε೽ϩޑᓓ᡺Ѻьวғӧ߈ъԃǴڬҶΒВܴᡉၨӭǴύᓓکఁᓓਔ໔և౜

ঢ়ॶǴѺьޣԃសӧ 21 ྃև౜ঢ়ॶǴӕՔԃសӧ 23 ྃև౜ঢ়ॶǴத่Քӕ ՉǴӕՔ܄ձϩѲ֡ϬǴܻ϶ኧໆࢂځдᜪࠠޑ 10 ७а΢ǹεӭኧޑᓓ᡺ѝ Ԗ೏Ѻь 1 ԛǴ࣬՟ޑ٬ҔޣѝԖ 1 ঁӅӕѺьၸޑᓓ᡺ۚӭǴӵკ 5.2Ƕ

ऩᙯඤԜၗ਑໣ࣁ٬Ҕޣᆶᓓ᡺ޑΒϡંତǴԜંତԖ 3928 ՉȐමჹᓓ᡺Ѻ ьޑ٬ҔޣΓኧȑک 2691 ӈȐᓓ᡺ঁኧȑǴऩۓကี౧ำࡋࣁ௡௢௡௭௘௥௢ ௘௡௧௜௥௘௦

௧௢௧௔௟ ௘௡௧௥௜௘௦ ǴԜၗ

਑໣ޑี౧ำࡋࣁ 0.00078Ƕ

5.2 ᒧ ᒧ᏷࣬՟ࡋෳໆݤ

ӧҁس಍ύǴѝा࣬՟ࡋෳໆݤளډޑ่݀ᆶ΋٤୷ҁྗ߾ߥ࡭΋ठǴךॺ ൩ёаԾҗᒧ᏷ૈౢғനӳ่݀ޑ࣬՟ࡋෳໆݤǹӣ៝ക࿯ 3.3.2ǴךॺׯቪΑΟ ᅿ࣬՟ࡋෳໆݤǴ٬ځԋࣁБำԄ 3.2ǵБำԄ 3.3 کБำԄ 3.4Ƕ

ӧᒧ᏷࣬՟ࡋෳໆݤϐ߻ǴךॺёаӃڰۓҔܭᑔᒧংᒧᓓ᡺ޑຯᚆୖኧǴ ӢࣁΓॺ೯தό཮Եቾϼᇻޑᓓ᡺ǴӚຯᚆୖኧ܌఼ᇂޑጄൎӵკ 5.3Ǻ

კ 5.3 Ӛຯᚆୖኧ܌఼ᇂޑጄൎȐаѠчًઠࣁύЈȑ

ךॺёаว౜྽ຯᚆୖኧ೛ࣁ 5 ϦٚࡕǴځ఼ᇂޑጄൎς࿶ၲډΑ᏾ঁѱ୔Ǵ

ࡺҁჴᡍஒаԜຯᚆୖኧᝩុ຾ՉǴځᎩຯᚆୖኧஒ੮ډനࡕ૸ፕǶךॺᔈҔ LOOCVȐLeave-one-out cross-validationȑӧԜ੿ჴШࣚၗ਑໣ύຑሽӚᅿ࣬՟ࡋෳ

ໆݤޑ܄ૈǴӵკ 5.4Ƕᡍ᛾ၸำύǴךॺ཮౽ନᡍ᛾ၗ਑ϐѺьჹ૽ግၗ਑ϐୃ

ӳຑϩޑቹៜǶ

კ 5.4 ҁፕЎ٬Ҕ LOOCV ޑᡍ᛾ၸำ

࣬՟ࡋෳໆݤޑຑሽ่݀ӵკ 5.5Ǻ

კ 5.5 Οᅿ࣬՟ࡋෳໆݤϐ௢ᙚྗዴࡋ

ךॺёа࣮ډΟᅿ࣬՟ࡋෳໆݤӧ໻ୖԵՏ࿼ᆶຯᚆୖኧΠޑ߄౜΋ठǴ೭

ࢂӢࣁ࣬՟ޑ٬ҔޣѝԖ 1 ঁӅӕѺьၸޑᓓ᡺ۚӭǴӵკ 5.2ǹࡺךॺᒧ᏷ᙁൂ

ޑ Distance-basedǴ٠යఈѬჹኧॶ௵གޑ੝܄Ǵૈ୼ӧ௃ნ࣬ᜢ܄ޑу៾΢Ǵள ډКၨᡉ๱ޑਏ݀Ƕ

྽س಍ௗԏډ٬Ҕޣፎ؃௃ნࡕǴ཮ӃճҔՏ࿼ᆶຯᚆୖኧၸᘠёૈޑংᒧ ᓓ᡺Ǵךॺϩձஒҁس಍ޑ௢ᙚኧໆڰۓࣁ 10ǵ20 ک 30 ৎংᒧᓓ᡺Ǵຯᚆୖኧ ڰۓࣁ 500 ϦЁǵ1ǵ2ǵ3ǵ4 ک 5 ϦٚǴჹচۈ่݀کᑔᒧ௢ᙚኧໆλܭ၀ຯᚆ

ୖኧϐᓓ᡺ኧໆޑ௃ݩǴӚձ຾Չ໻ୖԵՏ࿼ᆶຯᚆୖኧޑ߃؁ຑሽǴӵკ 5.6 ک კ 5.7Ǻ

Distance-based Correlation-based Cosine-based

10 Candidates 15.24% 15.24% 15.24%

20 Candidates 20.50% 20.50% 20.50%

30 Candidates 24.62% 24.62% 24.62%

0%

5%

10%

15%

20%

25%

30%

Accuracy (%)

კ 5.6 ໻ୖԵՏ࿼ᆶຯᚆୖኧޑӚຯᚆୖኧϐྗዴࡋ

კ 5.7 ໻ୖԵՏ࿼ᆶຯᚆୖኧޑӚຯᚆୖኧϐྗዴࡋȐᑔᒧ௢ᙚኧໆλܭ၀ຯᚆୖኧϐᓓ᡺ኧໆޑ

௃ݩȑ

0 km 0.5 km 1 km 2 km 3 km 4 km 5 km Unlimited

10 candidates 85.59% 55.14% 38.15% 25.13% 19.03% 16.89% 15.24% 3.18%

20 candidates 85.59% 69.73% 50.79% 35.36% 26.88% 23.07% 20.50% 4.82%

30 candidates 85.59% 75.99% 59.67% 42.45% 32.48% 28.28% 24.62% 5.70%

0%

Accuracy (%)

0.5 km 1 km 2 km 3 km 4 km 5 km

10 candidates 43.14% 30.98% 20.54% 15.15% 13.15% 12.13%

20 candidates 57.90% 39.13% 28.93% 20.86% 17.75% 15.98%

30 candidates 67.01% 48.34% 34.26% 25.42% 21.73% 18.22%

0%

Accuracy (%)

ჴᡍ่݀ᡉҢǴຯᚆୖኧቚεਔǴ௢ᙚྗዴࡋሀ෧ǹ೭ࢂӢࣁ٬Ҕޣёᒧ᏷

Coefficient Age

0

Coefficient Sex Index

21.5

Coefficient Time of Day

24.8

Coefficient Weekday

0

Coefficient Month

0

Coefficient Number of Companion

0

Coefficient Alone-Friend

24.8

Coefficient Alone-Classmate

25.25

Coefficient Alone-Colleague

25.2

Coefficient Alone-Family

25.21

Coefficient Alone-Lover

25

Coefficient Friend-Classmate

24.7

Coefficient Friend-Colleague

25.2

Coefficient Friend-Family

25.05

Coefficient Friend-Lover

25.05

Coefficient Classmate-Colleague

25.2

Coefficient Classmate-Family

25.23

Coefficient Classmate-Lover

კ 5.8 ௃ნ࣬ᜢ܄ෳໆኳࠠϐӚ໨߯ኧॶޑѳ֡ڮύՏ࿼Ȑआᗺࣁന٫߯ኧȑ

Coefficient Colleague-Family

25.289

Coefficient Colleague-Lover

25.276

Coefficient Family-Lover

࡯࢕ࢋࢌ_ࢇ ࡯࢕ࢋࢌ_࢙ ࡯࢕ࢋࢌ_࢚ ࡯࢕ࢋࢌ_࢝ ࡯࢕ࢋࢌ_࢓ ࡯࢕ࢋࢌ_࢔ࢉ

1 1 0.1 0.4 1 1

Alone Friend Classmate Colleague Family Lover

Alone 1 1 0.8 1 1 0

Friend 1 1 1 0 1 0

Classmate 0.8 1 1 0 1 0

Colleague 1 0 0 1 1 0

Family 1 1 1 1 1 0

Lover 0 0 0 0 0 1

߄ 5.4 ᕴ่௃ნ࣬ᜢ܄ෳໆኳࠠޑӚ໨ന٫߯ኧ

5.4 ຑ ຑሽ

ӵ݀٬Ҕޣ໻ჹ΋λ೽ϩޑ໨ҞຑϩǴٗሶ௢ᙚྗዴࡋեόж߄س಍߄౜ό ӳǹ೭ࢂӢࣁس಍௢ᙚΑࡐӭ٬ҔޣؒԖຑϩޑ໨ҞǴࡐёૈдॺচҁ൩ߚத഻

៿೭٤໨Ҟ[24]Ƕ

ྗዴࡋଯޑ௢ᙚس಍Ψόૈߥ᛾٬Ҕޣჹ௢ᙚ่݀ᅈཀǹӵ݀س಍௢ᙚࢬՉ

໨Ҟ๏٬ҔޣǴε൯ගଯྗዴࡋǴՠ٬ҔޣࡐёૈԐςޕၰ೭٤ၗૻǴӢԜ٬Ҕ ޣό཮ᇡࣁ೭ኬޑ௢ᙚس಍ࢂԖሽॶޑǹ΋૓ԶقǴس಍௢ᙚߚࢬՉ໨Ҟ཮फ़ե

ྗዴࡋǴՠ٬ҔޣϸԶ৒ܰว౜΋٤ཥڻǵԾρפόډޑୃӳ໨Ҟ[24]Ƕ

ӢԜҁس಍аڐӕၸᘠݤࣁ୷ᘵǴவঁձ٬Ҕޣୃӳᕴ่იᡏୃӳǴ٠ᙖҗ ԏ໣ډޑޗҬკᆶ௃ნၗૻቚ຾س಍܄ૈǴයఈౢғᇯЬᅰ៿ޑᓓ᡺௢ᙚ่݀ǹ Զڐӕၸᘠޑ୏ᐒٰԾܭΓॺ೯த཮வࠔښ࣬՟ޑΓளډനӳޑࡌ᝼[25]Ƕ

നࡕךஒᔈҔཛྷ൨ډޑന٫߯ኧک LOOCV ӧ੿ჴШࣚၗ਑໣ύຑሽس಍܄

ૈǹࣁΑКၨ؂ঁ௃ნᆢࡋޑቹៜǴך౽ନόቹៜ௢ᙚޑ௃ნᆢࡋǴᢀჸঁձ܈

იᡏ٬Ҕޣ࣬՟ࡋεܭ 0 Զڙ௃ნቹៜޑ໨ҞȐऊэᕴኧ 1.76%ȑǴ଺Α 5 ፺ෳ၂Ǵ ӵ߄ 5.5Ǻ

߄ 5.5 Ӛ௃ნᆢࡋޑቹៜำࡋ

ӵ߄ 5.5 ܌ҢǴԵቾঁձ௃ნᗨฅ࣮՟Ԗ܌ׯ຾Ǵՠځ T-Test ϐ P Value ࠅᆶ Round 1 ৡόӭǹόၸ྽ךॺ຾΋؁Եቾ܌Ԗ௃ნࡕǴջࣁڙ௃ნቹៜޑ໨Ҟ஥ٰ

Αᡉ๱ޑׯ຾ǴT-Test ϐ P Value ᛾ჴځࣁᡉ๱НྗȐSignificant LevelȑǶ

നࡕǴךॺϩձஒҁس಍ޑ௢ᙚኧໆڰۓࣁ 10ǵ20 ک 30 ৎংᒧᓓ᡺Ǵຯᚆ

ୖኧڰۓࣁ 500 ϦЁǵ1ǵ2ǵ3ǵ4 ک 5 ϦٚǴᆶ୷ܭ໶ኧک୷ܭࢬՉ܄௢ᙚȐ೏

Ѻьԛኧӭޑᓓ᡺ᓬӃ௢ᙚȑޑБݤ຾ՉКၨǴ٠ჹচۈ่݀کᑔᒧ௢ᙚኧໆλ ܭ၀ຯᚆୖኧϐᓓ᡺ኧໆޑ௃ݩǴӚձ຾ՉຑሽǴӵკ 5.9 کკ 5.10ǶࣁΑ׳৒ܰ

Кၨس಍܄ૈǴךॺीᆉΑҁس಍ჹ୷ܭࢬՉ܄௢ᙚޑྗዴࡋԋߏ౗Ǵӵკ 5.11Ƕ Average Position T-Test (P Value) Round 1: Non-contextual information 25.29 1

Round 2: Time of Day 22.02 0.11 Round 3: Weekday 24.98 0.89 Round 4: Type of Companion 24.18 0.62

Round 5: All 21.22 0.04

კ 5.9 ӚБݤᆶຯᚆୖኧϐ௢ᙚྗዴࡋ

კ 5.10 ӚБݤᆶຯᚆୖኧϐ௢ᙚྗዴࡋȐᑔᒧ௢ᙚኧໆλܭ၀ຯᚆୖኧϐᓓ᡺ኧໆޑ௃ݩȑ

500 m 1 km 2 km 3 km 4 km 5 km

10 candidates (Random) 47.92% 30.55% 17.72% 13.42% 10.95% 9.28%

10 candidates (Popularity) 55.52% 37.64% 24.19% 15.40% 13.68% 12.27%

10 candidates (CF) 55.14% 38.15% 25.13% 19.03% 16.89% 15.24%

10 candidates (Context-CF) 55.21% 38.17% 25.14% 19.26% 16.96% 15.43%

20 candidates (Random) 62.06% 43.25% 26.63% 19.24% 16.08% 13.87%

20 candidates (Popularity) 69.71% 51.24% 34.36% 24.67% 18.68% 16.93%

20 candidates (CF) 69.73% 50.79% 35.36% 26.88% 23.07% 20.50%

20 candidates (Context-CF) 69.74% 50.84% 35.37% 27.00% 23.27% 20.76%

30 candidates (Random) 69.87% 50.85% 33.74% 24.39% 20.37% 17.32%

30 candidates (Popularity) 75.73% 59.42% 41.31% 31.25% 25.65% 20.03%

30 candidates (CF) 75.99% 59.67% 42.45% 32.48% 28.28% 24.62%

30 candidates (Context-CF) 75.99% 59.68% 42.51% 32.68% 28.32% 24.77%

0%

Accuracy (%)

500 m 1 km 2 km 3 km 4 km 5 km

10 candidates (Random) 31.46% 21.41% 12.07% 8.90% 6.59% 5.63%

10 candidates (Popularity) 43.79% 30.37% 19.46% 11.11% 9.60% 8.89%

10 candidates (CF) 43.14% 30.98% 20.54% 15.15% 13.15% 12.13%

10 candidates (Context-CF) 43.27% 31.00% 20.56% 15.40% 13.21% 12.34%

20 candidates (Random) 40.83% 27.72% 18.23% 11.90% 9.84% 8.56%

20 candidates (Popularity) 57.87% 39.80% 27.70% 18.26% 12.81% 12.00%

20 candidates (CF) 57.90% 39.13% 28.93% 20.86% 17.75% 15.98%

20 candidates (Context-CF) 57.93% 39.20% 28.95% 21.00% 17.98% 16.28%

30 candidates (Random) 50.11% 33.43% 22.80% 15.55% 12.31% 9.65%

30 candidates (Popularity) 66.31% 47.92% 32.75% 23.93% 18.61% 12.82%

30 candidates (CF) 67.01% 48.34% 34.26% 25.42% 21.73% 18.22%

30 candidates (Context-CF) 67.01% 48.36% 34.34% 25.66% 21.79% 18.40%

0%

Accuracy (%)

კ 5.11 ҁس಍ჹ୷ܭࢬՉ܄௢ᙚޑྗዴࡋԋߏ౗Ȑᑔᒧ௢ᙚኧໆλܭ၀ຯᚆୖኧϐᓓ᡺ኧໆޑ௃

ݩȑ

ӵ߻܌ॊǴ΋૓ԶقǴس಍௢ᙚߚࢬՉ໨Ҟ཮फ़եྗዴࡋǹӵკ 5.10 کკ 5.11

܌ҢǴ྽ҁس಍௢ᙚ 10 ৎংᒧᓓ᡺Ǵӧຯᚆ 500 ϦЁȐ؁Չऊ 10 ϩដȑޑ௃ნ ΠǴྗዴࡋࣁ 43.27%ǹౣեܭ୷ܭࢬՉ܄௢ᙚϐྗዴࡋ 43.79%Ǵԋߏ౗ࣁ-1.2%Ƕ ӧຯᚆ 1 ϦٚȐ؁Չऊ 20 ϩដȑޑ௃ნΠǴྗዴࡋࣁ 31%ǹ໒ۈౣଯܭ୷ܭࢬՉ

܄௢ᙚϐྗዴࡋ 30.37%Ǵԋߏ౗ࣁ 2.08%Ƕӧຯᚆ 5 Ϧٚޑ௃ნΠȐམϦًऊ 30 ϩដȑǴྗዴࡋࣁ 12.34%ǹଯܭ୷ܭࢬՉ܄௢ᙚϐྗዴࡋ 8.89%Ǵԋߏ౗ࣁ 38.8%Ƕ ೭ཀښ๱Ǵӵ݀٬Ҕޣ൨פᓓ᡺܌೛ۓޑጄൎКၨεǴ࣬ၨܭ୷ܭࢬՉ܄௢ᙚǴ ҁس಍ёаౢғ׳ӳޑ௢ᙚ่݀Ƕ

500 m 1 km 2 km 3 km 4 km 5 km

10 candidates -1.20% 2.08% 5.66% 38.66% 37.59% 38.80%

20 candidates 33.74% -1.49% 4.51% 15.01% 40.30% 35.66%

30 candidates 1.05% 0.92% 4.83% 7.24% 17.11% 43.46%

-5.00%

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%

50.00%

Growth rate (%)

6 ่ ่ፕ

ҁፕЎගр٬ҔޗҬკᆶ௃ნགޕϐՉ୏ᓓ᡺௢ᙚس಍ǹಃ 3 കךॺ೸ၸ Graph API ڗளޗҬკᆶ໒ܫკύǴ٬Ҕޣᆶځܻ϶ჹᓓ᡺ޑୃӳک܌Ԗёૈቹៜ Γॺᒧ᏷ᓓ᡺ޑӢનǴךॺஒ೭٤ӢનᆀࣁޗҬკᆶ௃ნၗૻǴ٠೸ၸس಍ࢎᄬǵ

ၗ਑ኳࠠǵБำԄᆶᄽᆉݤϟಏӵՖஒځ᏾ӝډ௢ᙚس಍ύǹಃ 4 ക௶ॊჴբמ

ೌᆶࢬำǴ٠৖Ң٬Ҕޣϟय़ǹಃ 5 കޑჴᡍ่݀ᡉҢΑҁس಍ڑຫޑ܄ૈǶ ӧҁകύǴക࿯ 6.1 ஒᕴ่ךॺ܌଺ޑπբکளډޑԋ݀Ǵക࿯ 6.2 ᇥܴس಍

҂ٰว৖ᆶࣴزБӛǶ

6.1 ᕴ่ଅ᝘

ҁፕЎճҔ Facebook ໒ܫკȐOpen GraphȑޑѺьၗ਑ȐCheck-insȑ೛ी΋

ঁՉ୏ᓓ᡺௢ᙚس಍Ǵ٠ჴբډ Windows Azure Platform ΢Ƕس಍аڐӕၸᘠݤ ȐCollaborative Filteringȑࣁ୷ᘵǴவঁձ٬Ҕޣୃӳᕴ่იᡏୃӳǴჴ౜იᡏ௢

ᙚ୍ܺǹ٠ԵቾޗҬკȐSocial Graphȑᆶ௃ნၗૻȐContextual Informationȑගϲ

௢ᙚࠔ፦

΋૓ԶقǴس಍௢ᙚߚࢬՉ໨Ҟ཮फ़եྗዴࡋǹჴᡍ่݀ᡉҢǴҁس಍ӧύǵ ߏຯᚆȐ3 ډ 5 Ϧٚȑޑ௃ნΠǴྗዴࡋ࣬ၨܭ୷ܭࢬՉ܄௢ᙚԖᡉ๱ԋߏǴԋߏ

౗ऊ 38%Ƕ೭ཀښ๱Ǵӵ݀٬Ҕޣ൨פᓓ᡺܌೛ۓޑጄൎКၨεǴ࣬ၨܭ୷ܭࢬ Չ܄௢ᙚǴҁس಍ёаౢғ׳ӳޑ௢ᙚ่݀Ƕ

6.2 ҂ ҂ٰπբ

ӵ߄ 5.1 ک߄ 5.3 ܌ҢǴ1 Տ٬Ҕޣѳ֡Ԗ 504 Տܻ϶ǴԌନख़ፄޑჹຝǴऊ ё஥ٰ 365 Տ٬Ҕޣޑၗ਑Ȑԏ໣ډޑ٬ҔޣΓኧ/ڙෳޣΓኧȑǴӢԜҁس಍ᒿ๱

٬ҔޣቚуǴёୖԵޑ࣬՟٬Ҕޣኧໆ཮ε൯ԋߏǴஒԖշܭ຾΋؁ගϲ௢ᙚࠔ

፦ǴՠځीᆉໆΨஒԋጕ܄уεǹჹܭՉ୏း࿼ٰᇥǴӣᔈೲࡋࢂቹៜ٬Ҕޣᡏ ᡍനख़ाޑӢનϐ΋ǴӢԜӵՖׯ๓ځёᘉ৖܄ȐScalabilityȑаϷӵՖᆢៈঁΓ ᗦدȐPrivacyȑࢂॶளుΕࣴزޑ᝼ᚒǶ

ୖ

ୖԵЎ᝘

1. 㜤ࡌ㡚, ڬ㫓, and ؋ޚֻ, 㚚܄ϯ௢૚س䶘ޑࣴز僳৖. Ծฅࣽ䗄僳৖, 2009. 19(001): p. 1-15.

2. Schilit, B.N. and M.M. Theimer, Disseminating active map information to mobile hosts. Network, IEEE, 1994. 8(5): p. 22-32.

3. Dey, A.K., Understanding and using context. Personal and ubiquitous computing, 2001. 5(1): p. 4-7.

4. Woerndl, W. and J. Schlichter. Introducing context into recommender systems. in Proceedings of AAAI 2007 Workshop on Recommender Systems in e-Commerce.

2007.

5. Wikipedia contributors. Social graph. Available from:

http://en.wikipedia.org/w/index.php?title=Social_graph&oldid=495500805.

6. Facebook. Open Graph. Available from:

http://developers.facebook.com/docs/opengraph/.

7. Wikipedia contributors. Check-in. Available from:

http://en.wikipedia.org/w/index.php?title=Check-in&oldid=495397467.

8. Facebook. Graph API. Available from:

http://developers.facebook.com/docs/reference/api/.

9. Goldberg, D., et al., Using collaborative filtering to weave an information tapestry. Communications of the ACM, 1992. 35(12): p. 61-70.

10. Konstan, J.A., et al., GroupLens: applying collaborative filtering to Usenet news.

Communications of the ACM, 1997. 40(3): p. 77-87.

11. Adomavicius, G. and A. Tuzhilin, Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions. Knowledge and Data Engineering, IEEE Transactions on, 2005. 17(6): p. 734-749.

12. Sarwar, B., et al. Item-based collaborative filtering recommendation algorithms.

in Proceedings of the 10th international conference on World Wide Web. 2001.

ACM.

13. Claypool, M., et al. Combining content-based and collaborative filters in an online newspaper. in Proceedings of ACM SIGIR Workshop on Recommender Systems. 1999. Citeseer.

14. Pazzani, M.J., A framework for collaborative, content-based and demographic filtering. Artificial Intelligence Review, 1999. 13(5): p. 393-408.

15. %DODEDQRYLü0DQG<6KRKDPFab: content-based, collaborative recommendation. Communications of the ACM, 1997. 40(3): p. 66-72.

16. Adomavicius, G., et al., Incorporating contextual information in recommender systems using a multidimensional approach. ACM Transactions on Information Systems (TOIS), 2005. 23(1): p. 103-145.

17. Chen, A., Context-aware collaborative filtering system: Predicting the user’s preference in the ubiquitous computing environment. Location-and

Context-Awareness, 2005: p. 75-81.

18. Nguyen, Q.N. and F. Ricci. Long-term and session-specific user preferences in a mobile recommender system. in Proceedings of the 13th international

conference on Intelligent user interfaces. 2008. ACM.

19. Sadeh, N., E. Chan, and L. Van. MyCampus: an agent-based environment for context-aware mobile services. in Proceedings of Workshop on Ubiquitous Agents on embedded, wearable and mobile devices. 2002.

20. ໳௴჏, ௃ნၗૻჹඵችࠠး࿼΢ᓓ᡺௢ᙚس಍ޑቹៜϩ݋, in ᆵ᡼εᏢ

ၗૻπำᏢࣴز܌ᏢՏፕЎ2009, ᆵ᡼εᏢ.

21. Park, M.H., H.S. Park, and S.B. Cho. Restaurant recommendation for group of people in mobile environments using probabilistic multi-criteria decision making.

in Proceedings of the 8th Asia-3DFL¿FFRQIHUHQFHRQ&RPSXWHU-Human Interaction. 2008. Springer.

22. Wikipedia contributors. Cloud computing. Available from:

http://en.wikipedia.org/w/index.php?title=Cloud_computing&oldid=499416499.

23. Sarwar, B., et al. Analysis of recommendation algorithms for e-commerce. in Proceedings of the 2nd ACM conference on Electronic commerce. 2000. ACM.

24. 㜤ࡌ㡚, et al., 㚚܄ϯ௢૚س䶘侶ᣴБݤ䶵ॊ. ᯕ䩮س䶘ᢳᯕ䩮܄ࣽ䗄, 2009. 6(003): p. 1-10.

25. Wikipedia contributors. Collaborative filtering. Available from:

http://en.wikipedia.org/w/index.php?title=Collaborative_filtering&oldid=49550 4334.

ߕ ߕᒵ

໨Ҟ К౗

මѺьޑ٬ҔޣتζК(ت/ζ) 1.02 3719 / 3654

මѺьޑ٬ҔޣК౗ 29.36% 7395 / 25184

ѺьӧεѠчӦ୔ޑК౗ 63.64% 47277 / 74283 ӦᗺӧεѠчӦ୔ޑК౗ 49.74% 18556 / 37305 මჹεѠчӦ୔ӦᗺѺьޑ٬ҔޣК౗ 76.50% 5657 / 7395

ߕᒵ 1 ԏ໣ډޑӚ໨ၗ਑ϐК౗Ȑ܌ԖӦᗺȑ

ߕᒵ 2 ԏ໣ډޑ܌ԖӦᗺϐϩѲკ

໨

໨Ҟ ѳѳ֡ॶ ύύՏኧ ኱኱ྗৡ നനεॶ നനλॶ

මѺьޑ٬Ҕޣԃស 22.91 22 4.56 106 13 n=7395

٬ҔޣޑѺьԛኧ 3.27 0 6.61 40 0 n=25184

٬ҔޣѺьၸޑӦᗺ

ঁኧ

2.95 0 5.99 39 0 n=25184 Ӧᗺ೏Ѻьޑԛኧ 2.21 1 6.00 256 1 n=37305 මѺьޑ٬ҔޣѺь

ԛኧ

11.14 11 7.83 40 1 n=7395 මѺьޑ٬ҔޣѺь

ၸޑӦᗺঁኧ

10.05 10 7.13 39 1 n=7395

܌ԖѺьޑѺьޣԃ ស

23.05 22 4.50 106 13 n=74283

܌ԖѺьޑӦᗺ೏Ѻ ьԛኧ

1.11 1 0.53 19 1 n=74283

܌ԖѺьޑӕՔѳ֡

ԃស

23.68 23 4.58 107 14 n=74283

܌ԖѺьޑӕՔ܄ձ

ࡰ኱(ζ/ᕴኧ)

0.51 0.5 0.49 1 0 n=74283

܌ԖѺьޑӕՔΓኧ 2.36 2 2.42 51 1 n=74283

ߕᒵ 3 ԏ໣ډޑӚ໨ၗ਑ϐѳ֡ॶǵύՏኧᆶ኱ྗৡȐ܌ԖӦᗺȑ

0

2010/8 2010/10 2010/12 2011/2 2011/4 2011/6 2011/8 2011/10 2011/12 2012/2 2012/4

܌ԖѺьኧໆ

ߕᒵ 4 ಍ीԏ໣ډޑ܌ԖѺьၗ਑

在文檔中 使用社交圖與情境感知之行動餐廳推薦系統 (頁 34-65)