ݺߣshows by User: domonkostikk / http://www.slideshare.net/images/logo.gif ݺߣshows by User: domonkostikk / Thu, 24 Mar 2016 09:42:57 GMT ݺߣShare feed for ݺߣshows by User: domonkostikk Lessons learnt at building recommendation services at industry scale /slideshow/lessons-learnt-at-building-recommendation-services-at-industry-scale/59982420 gravitylessonslearntatbuildingrecservicesatindustryscalefinal-160324094257
Industry day keynote presentation held at ECIR 2016, Padova. The talk presents algorithmic, technical and business challenges Gravity R&D encountered from building a recommender system vendor company from being a top Netflix Prize contender.]]>
Industry day keynote presentation held at ECIR 2016, Padova. The talk presents algorithmic, technical and business challenges Gravity R&D encountered from building a recommender system vendor company from being a top Netflix Prize contender.]]> Thu, 24 Mar 2016 09:42:57 GMT /slideshow/lessons-learnt-at-building-recommendation-services-at-industry-scale/59982420 domonkostikk@slideshare.net(domonkostikk) Lessons learnt at building recommendation services at industry scale domonkostikk Industry day keynote presentation held at ECIR 2016, Padova. The talk presents algorithmic, technical and business challenges Gravity R&D encountered from building a recommender system vendor company from being a top Netflix Prize contender. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/gravitylessonslearntatbuildingrecservicesatindustryscalefinal-160324094257-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Industry day keynote presentation held at ECIR 2016, Padova. The talk presents algorithmic, technical and business challenges Gravity R&amp;D encountered from building a recommender system vendor company from being a top Netflix Prize contender.
Lessons learnt at building recommendation services at industry scale from Domonkos Tikk
]]> 3048 8 https://cdn.slidesharecdn.com/ss_thumbnails/gravitylessonslearntatbuildingrecservicesatindustryscalefinal-160324094257-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Recommenders on video sharing portals - business and algorithmic aspects /slideshow/recommenders-on-video-sharing-portals-business-and-algorithmic-aspects/53789135 gravitymedianet2015-151011085335-lva1-app6891
This talk was given at HTE Medianet 2015 conference on the business ecosystem and its impact and requirement on recommender solutions on user generated content (UGC) video sharing portals (like Youtube, Dailymotion, etc). Besides the intro on Gravity R&D and the business model, we present some information on a real-world case study. We show that larger view counts are directly influence the recommendation quality.]]>
This talk was given at HTE Medianet 2015 conference on the business ecosystem and its impact and requirement on recommender solutions on user generated content (UGC) video sharing portals (like Youtube, Dailymotion, etc). Besides the intro on Gravity R&D and the business model, we present some information on a real-world case study. We show that larger view counts are directly influence the recommendation quality.]]> Sun, 11 Oct 2015 08:53:35 GMT /slideshow/recommenders-on-video-sharing-portals-business-and-algorithmic-aspects/53789135 domonkostikk@slideshare.net(domonkostikk) Recommenders on video sharing portals - business and algorithmic aspects domonkostikk This talk was given at HTE Medianet 2015 conference on the business ecosystem and its impact and requirement on recommender solutions on user generated content (UGC) video sharing portals (like Youtube, Dailymotion, etc). Besides the intro on Gravity R&D and the business model, we present some information on a real-world case study. We show that larger view counts are directly influence the recommendation quality. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/gravitymedianet2015-151011085335-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> This talk was given at HTE Medianet 2015 conference on the business ecosystem and its impact and requirement on recommender solutions on user generated content (UGC) video sharing portals (like Youtube, Dailymotion, etc). Besides the intro on Gravity R&amp;D and the business model, we present some information on a real-world case study. We show that larger view counts are directly influence the recommendation quality.
Recommenders on video sharing portals - business and algorithmic aspects from Domonkos Tikk
]]> 919 7 https://cdn.slidesharecdn.com/ss_thumbnails/gravitymedianet2015-151011085335-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Neighbor methods vs matrix factorization - case studies of real-life recommendations (Gravity LSRS2015 RECSYS 2015) /slideshow/neighbor-methods-vs-matrix-factorization-case-studies-of-reallife-recommendations-gravity-lsrs2015-recsys-2015/53051002 gravitylsrs2015recsysneighbourmethodsvsmf-150922080707-lva1-app6891
This talk was given by István Pilászy, co-founder and head of core development at Gravity R&D, at LSRS workshop at Recsys 2015. Messages of the talk: (1) in industry item-2-item (i2i) recommendation is the dominant case, hardly researched by academia; (2) in industry you have typically implicit feedback data; (3) matrix factorization (MF) is good to optimize error metric, but less obvious for top-N and i2i recommendations. (4) item-kNN in most cases outperforms MF for i2i in terms of CTR; (5) Performance heavily depends on the domain and the recommendation scenario.]]>
This talk was given by István Pilászy, co-founder and head of core development at Gravity R&D, at LSRS workshop at Recsys 2015. Messages of the talk: (1) in industry item-2-item (i2i) recommendation is the dominant case, hardly researched by academia; (2) in industry you have typically implicit feedback data; (3) matrix factorization (MF) is good to optimize error metric, but less obvious for top-N and i2i recommendations. (4) item-kNN in most cases outperforms MF for i2i in terms of CTR; (5) Performance heavily depends on the domain and the recommendation scenario.]]> Tue, 22 Sep 2015 08:07:07 GMT /slideshow/neighbor-methods-vs-matrix-factorization-case-studies-of-reallife-recommendations-gravity-lsrs2015-recsys-2015/53051002 domonkostikk@slideshare.net(domonkostikk) Neighbor methods vs matrix factorization - case studies of real-life recommendations (Gravity LSRS2015 RECSYS 2015) domonkostikk This talk was given by István Pilászy, co-founder and head of core development at Gravity R&D, at LSRS workshop at Recsys 2015. Messages of the talk: (1) in industry item-2-item (i2i) recommendation is the dominant case, hardly researched by academia; (2) in industry you have typically implicit feedback data; (3) matrix factorization (MF) is good to optimize error metric, but less obvious for top-N and i2i recommendations. (4) item-kNN in most cases outperforms MF for i2i in terms of CTR; (5) Performance heavily depends on the domain and the recommendation scenario. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/gravitylsrs2015recsysneighbourmethodsvsmf-150922080707-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> This talk was given by István Pilászy, co-founder and head of core development at Gravity R&amp;D, at LSRS workshop at Recsys 2015. Messages of the talk: (1) in industry item-2-item (i2i) recommendation is the dominant case, hardly researched by academia; (2) in industry you have typically implicit feedback data; (3) matrix factorization (MF) is good to optimize error metric, but less obvious for top-N and i2i recommendations. (4) item-kNN in most cases outperforms MF for i2i in terms of CTR; (5) Performance heavily depends on the domain and the recommendation scenario.
Neighbor methods vs matrix factorization - case studies of real-life recommendations (Gravity LSRS2015 RECSYS 2015) from Domonkos Tikk
]]> 5827 15 https://cdn.slidesharecdn.com/ss_thumbnails/gravitylsrs2015recsysneighbourmethodsvsmf-150922080707-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Challenges Encountered by Scaling Up Recommendation Services at Gravity R&D /slideshow/challenges-encountered-by-scaling-up-recommendation-services-at-gravity-rd/52993473 gravity-industry-session-recsys2015-150920215856-lva1-app6891
This talk was given by Bottyan Németh (Gravity R&D Product Owner & Co-founder) in the industry session at ACM Recsys Conference 2015 in Vienna. Presentation describes the challenges and solution we encountered by scaling up the recommendation services provided by Gravity.]]>
This talk was given by Bottyan Németh (Gravity R&D Product Owner & Co-founder) in the industry session at ACM Recsys Conference 2015 in Vienna. Presentation describes the challenges and solution we encountered by scaling up the recommendation services provided by Gravity.]]> Sun, 20 Sep 2015 21:58:56 GMT /slideshow/challenges-encountered-by-scaling-up-recommendation-services-at-gravity-rd/52993473 domonkostikk@slideshare.net(domonkostikk) Challenges Encountered by Scaling Up Recommendation Services at Gravity R&D domonkostikk This talk was given by Bottyan Németh (Gravity R&D Product Owner & Co-founder) in the industry session at ACM Recsys Conference 2015 in Vienna. Presentation describes the challenges and solution we encountered by scaling up the recommendation services provided by Gravity. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/gravity-industry-session-recsys2015-150920215856-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> This talk was given by Bottyan Németh (Gravity R&amp;D Product Owner &amp; Co-founder) in the industry session at ACM Recsys Conference 2015 in Vienna. Presentation describes the challenges and solution we encountered by scaling up the recommendation services provided by Gravity.
Challenges Encountered by Scaling Up Recommendation Services at Gravity R&D from Domonkos Tikk
]]> 1672 5 https://cdn.slidesharecdn.com/ss_thumbnails/gravity-industry-session-recsys2015-150920215856-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 General factorization framework for context-aware recommendations /slideshow/general-factorization-framework-for-contextaware-recommendations/47984688 2cf0a954-a52f-4510-9230-12b7d6fc885b-150511080104-lva1-app6891
]]>
]]> Mon, 11 May 2015 08:01:03 GMT /slideshow/general-factorization-framework-for-contextaware-recommendations/47984688 domonkostikk@slideshare.net(domonkostikk) General factorization framework for context-aware recommendations domonkostikk <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/2cf0a954-a52f-4510-9230-12b7d6fc885b-150511080104-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br>
General factorization framework for context-aware recommendations from Domonkos Tikk
]]> 361 2 https://cdn.slidesharecdn.com/ss_thumbnails/2cf0a954-a52f-4510-9230-12b7d6fc885b-150511080104-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds document 000000 http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Tartalomgazdagítás (content enrichment) /slideshow/tartalomgazdagts-content-enrichment/42860041 2014-141219015333-conversion-gate01
This is the slides from conTEXT conference held by SPSS Consulting (Hungary) at 2014.11.21. ݺߣs are in Hungarian Tartalomgazdagítás szövegbányászattal Linked Open Data alapján és felhasználása ajánlórendszerekben Ajánlórendszerek jellemzően két adatforrás alapján dolgoznak: a felhasználók tartalomfogyasztási szokásaik (collaborative filtering) és a tartalomak metaadatai alapján (content based filtering). A tartalomleíró adatoknak nagy előnye, hogy az ajánlások könnyen magyarázhatóak a felhasználóknak (azért ajánjuk, mert szereted Brad Pitt filmjeit, stb.), és már új tartalmak esetén is alkalmazhatók. Ezért nyilván nagyon fontos, hogy jó minőségűek és minél gazdagabbak legyenek a tartalmakat leíró metaadatok. Ezek a feltételek azonban gyakran nem teljesülnek, ezért szükség ún. tartalomgazdagításra (content enrichment), hogy jobban jellemezhetőek legyenek a tartalmak. A Gravity alkalmazott megoldás nyilvánosan elérhető adatbázisok (Linked Open Data, LOD) segítségével végzi el a tartalomgazdagítást, mint pl. a Freebase és a DBpedia. A probléma számos szövegbányászati részfeladatot tartalmaz, mint pl. névelemek felismerése és egyértelműsítése (színész, rendező neve; film címe), névelemek tulajdonságainak meghatározása (pl. színész fontosabb adatai) különböző forrásokból származó adatok egyesítése, átfedő, ill. esetleg inkonzisztens adatok kanonizálása. Az előadás során röviden ismertetésre kerül az általunk alkalmazott RDF alapú megoldás, a SPARQL lekérdezőnyelv és alkalmazása, és rámutatunk néhány lehetséges megoldására a szövegbányázati feladatoknak.]]>
This is the slides from conTEXT conference held by SPSS Consulting (Hungary) at 2014.11.21. ݺߣs are in Hungarian Tartalomgazdagítás szövegbányászattal Linked Open Data alapján és felhasználása ajánlórendszerekben Ajánlórendszerek jellemzően két adatforrás alapján dolgoznak: a felhasználók tartalomfogyasztási szokásaik (collaborative filtering) és a tartalomak metaadatai alapján (content based filtering). A tartalomleíró adatoknak nagy előnye, hogy az ajánlások könnyen magyarázhatóak a felhasználóknak (azért ajánjuk, mert szereted Brad Pitt filmjeit, stb.), és már új tartalmak esetén is alkalmazhatók. Ezért nyilván nagyon fontos, hogy jó minőségűek és minél gazdagabbak legyenek a tartalmakat leíró metaadatok. Ezek a feltételek azonban gyakran nem teljesülnek, ezért szükség ún. tartalomgazdagításra (content enrichment), hogy jobban jellemezhetőek legyenek a tartalmak. A Gravity alkalmazott megoldás nyilvánosan elérhető adatbázisok (Linked Open Data, LOD) segítségével végzi el a tartalomgazdagítást, mint pl. a Freebase és a DBpedia. A probléma számos szövegbányászati részfeladatot tartalmaz, mint pl. névelemek felismerése és egyértelműsítése (színész, rendező neve; film címe), névelemek tulajdonságainak meghatározása (pl. színész fontosabb adatai) különböző forrásokból származó adatok egyesítése, átfedő, ill. esetleg inkonzisztens adatok kanonizálása. Az előadás során röviden ismertetésre kerül az általunk alkalmazott RDF alapú megoldás, a SPARQL lekérdezőnyelv és alkalmazása, és rámutatunk néhány lehetséges megoldására a szövegbányázati feladatoknak.]]> Fri, 19 Dec 2014 01:53:33 GMT /slideshow/tartalomgazdagts-content-enrichment/42860041 domonkostikk@slideshare.net(domonkostikk) Tartalomgazdagítás (content enrichment) domonkostikk This is the slides from conTEXT conference held by SPSS Consulting (Hungary) at 2014.11.21. ݺߣs are in Hungarian Tartalomgazdagítás szövegbányászattal Linked Open Data alapján és felhasználása ajánlórendszerekben Ajánlórendszerek jellemzően két adatforrás alapján dolgoznak: a felhasználók tartalomfogyasztási szokásaik (collaborative filtering) és a tartalomak metaadatai alapján (content based filtering). A tartalomleíró adatoknak nagy előnye, hogy az ajánlások könnyen magyarázhatóak a felhasználóknak (azért ajánjuk, mert szereted Brad Pitt filmjeit, stb.), és már új tartalmak esetén is alkalmazhatók. Ezért nyilván nagyon fontos, hogy jó minőségűek és minél gazdagabbak legyenek a tartalmakat leíró metaadatok. Ezek a feltételek azonban gyakran nem teljesülnek, ezért szükség ún. tartalomgazdagításra (content enrichment), hogy jobban jellemezhetőek legyenek a tartalmak. A Gravity alkalmazott megoldás nyilvánosan elérhető adatbázisok (Linked Open Data, LOD) segítségével végzi el a tartalomgazdagítást, mint pl. a Freebase és a DBpedia. A probléma számos szövegbányászati részfeladatot tartalmaz, mint pl. névelemek felismerése és egyértelműsítése (színész, rendező neve; film címe), névelemek tulajdonságainak meghatározása (pl. színész fontosabb adatai) különböző forrásokból származó adatok egyesítése, átfedő, ill. esetleg inkonzisztens adatok kanonizálása. Az előadás során röviden ismertetésre kerül az általunk alkalmazott RDF alapú megoldás, a SPARQL lekérdezőnyelv és alkalmazása, és rámutatunk néhány lehetséges megoldására a szövegbányázati feladatoknak. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/2014-141219015333-conversion-gate01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> This is the slides from conTEXT conference held by SPSS Consulting (Hungary) at 2014.11.21. ݺߣs are in Hungarian Tartalomgazdagítás szövegbányászattal Linked Open Data alapján és felhasználása ajánlórendszerekben Ajánlórendszerek jellemzően két adatforrás alapján dolgoznak: a felhasználók tartalomfogyasztási szokásaik (collaborative filtering) és a tartalomak metaadatai alapján (content based filtering). A tartalomleíró adatoknak nagy előnye, hogy az ajánlások könnyen magyarázhatóak a felhasználóknak (azért ajánjuk, mert szereted Brad Pitt filmjeit, stb.), és már új tartalmak esetén is alkalmazhatók. Ezért nyilván nagyon fontos, hogy jó minőségűek és minél gazdagabbak legyenek a tartalmakat leíró metaadatok. Ezek a feltételek azonban gyakran nem teljesülnek, ezért szükség ún. tartalomgazdagításra (content enrichment), hogy jobban jellemezhetőek legyenek a tartalmak. A Gravity alkalmazott megoldás nyilvánosan elérhető adatbázisok (Linked Open Data, LOD) segítségével végzi el a tartalomgazdagítást, mint pl. a Freebase és a DBpedia. A probléma számos szövegbányászati részfeladatot tartalmaz, mint pl. névelemek felismerése és egyértelműsítése (színész, rendező neve; film címe), névelemek tulajdonságainak meghatározása (pl. színész fontosabb adatai) különböző forrásokból származó adatok egyesítése, átfedő, ill. esetleg inkonzisztens adatok kanonizálása. Az előadás során röviden ismertetésre kerül az általunk alkalmazott RDF alapú megoldás, a SPARQL lekérdezőnyelv és alkalmazása, és rámutatunk néhány lehetséges megoldására a szövegbányázati feladatoknak.
Tartalomgazdag鱈t叩s (content enrichment) from Domonkos Tikk
]]> 677 1 https://cdn.slidesharecdn.com/ss_thumbnails/2014-141219015333-conversion-gate01-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Idomaar crowd rec_reference_fw /slideshow/idomaar-crowd-recreferencefw/40150455 idomaarcrowdrecreferencefw-141011141617-conversion-gate02
Idomaar is an open-source benchmarking framework for recommender systems created by the CrowdRec EU-project. It enables impartial evaluation of recommenders solutions from different aspect: (1) recommendation quality (2) technical aspects (3) business aspects. Main contributors of Idomaar: Moviri, Gravity RD, Technical University of Delft, Technical University Berlin]]>
Idomaar is an open-source benchmarking framework for recommender systems created by the CrowdRec EU-project. It enables impartial evaluation of recommenders solutions from different aspect: (1) recommendation quality (2) technical aspects (3) business aspects. Main contributors of Idomaar: Moviri, Gravity RD, Technical University of Delft, Technical University Berlin]]> Sat, 11 Oct 2014 14:16:17 GMT /slideshow/idomaar-crowd-recreferencefw/40150455 domonkostikk@slideshare.net(domonkostikk) Idomaar crowd rec_reference_fw domonkostikk Idomaar is an open-source benchmarking framework for recommender systems created by the CrowdRec EU-project. It enables impartial evaluation of recommenders solutions from different aspect: (1) recommendation quality (2) technical aspects (3) business aspects. Main contributors of Idomaar: Moviri, Gravity RD, Technical University of Delft, Technical University Berlin <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/idomaarcrowdrecreferencefw-141011141617-conversion-gate02-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Idomaar is an open-source benchmarking framework for recommender systems created by the CrowdRec EU-project. It enables impartial evaluation of recommenders solutions from different aspect: (1) recommendation quality (2) technical aspects (3) business aspects. Main contributors of Idomaar: Moviri, Gravity RD, Technical University of Delft, Technical University Berlin
Idomaar crowd rec_reference_fw from Domonkos Tikk
]]> 1200 2 https://cdn.slidesharecdn.com/ss_thumbnails/idomaarcrowdrecreferencefw-141011141617-conversion-gate02-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Big Data in Online Classifieds /slideshow/big-data-in-online-classifieds/35117797 tikkdomonkosicmaworkshop23052014-140526032943-phpapp01
The slideshow was presented at ICMA Conference in Helsinki at the "How to Turn Big Data into Dollars" Workshop organized by Gravity R&D, The presentation reviews the heterogeneity of data sources at classified media, shows the massive size of data available, and give some insights how to use those data for personalization in various scenarios.]]>
The slideshow was presented at ICMA Conference in Helsinki at the "How to Turn Big Data into Dollars" Workshop organized by Gravity R&D, The presentation reviews the heterogeneity of data sources at classified media, shows the massive size of data available, and give some insights how to use those data for personalization in various scenarios.]]> Mon, 26 May 2014 03:29:43 GMT /slideshow/big-data-in-online-classifieds/35117797 domonkostikk@slideshare.net(domonkostikk) Big Data in Online Classifieds domonkostikk The slideshow was presented at ICMA Conference in Helsinki at the "How to Turn Big Data into Dollars" Workshop organized by Gravity R&D, The presentation reviews the heterogeneity of data sources at classified media, shows the massive size of data available, and give some insights how to use those data for personalization in various scenarios. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/tikkdomonkosicmaworkshop23052014-140526032943-phpapp01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> The slideshow was presented at ICMA Conference in Helsinki at the &quot;How to Turn Big Data into Dollars&quot; Workshop organized by Gravity R&amp;D, The presentation reviews the heterogeneity of data sources at classified media, shows the massive size of data available, and give some insights how to use those data for personalization in various scenarios.
Big Data in Online Classifieds from Domonkos Tikk
]]> 2082 3 https://cdn.slidesharecdn.com/ss_thumbnails/tikkdomonkosicmaworkshop23052014-140526032943-phpapp01-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Context-aware similarities within the factorization framework - presented at CARR 2013 /slideshow/contextaware-similarities-within-the-factorization-framework/16523070 casimcarr13-130214032439-phpapp02
Talk deals with a practical recommendation sceaario - item-2-item recommendation (similar/related items) with implicit feedback and context. Solution is provided in the factorization framework. Paper abstract (to appear in ACM digital Library) Item-to-item recommendation - when the most similar items sought to the actual item - is an important recommendation scenario in practical recommender systems. One way to solve this task is to use the similarity between item feature vectors of factorization models. By doing so, one may transfer the well-known accuracy of factorization models observed at the personalized recommendations to the item-to-item case. This paper introduces context-awareness to item similarities in the factorization framework. Two levels of context-aware similarities are defined and applied to two context-aware implicit feedback based factorization methods (iTALS and iTALSx). We investigate the advantages and drawbacks of the approaches on four real life implicit feedback data sets and we characterize the conditions for their application. The results suggest that it is worth using contextual information for item-to-item recommendations in the factorization framework, however, one should carefully select the appropriate method to achieve similar accuracy gain than in the case of the more general item-to-user recommendation scenario. ]]>
Talk deals with a practical recommendation sceaario - item-2-item recommendation (similar/related items) with implicit feedback and context. Solution is provided in the factorization framework. Paper abstract (to appear in ACM digital Library) Item-to-item recommendation - when the most similar items sought to the actual item - is an important recommendation scenario in practical recommender systems. One way to solve this task is to use the similarity between item feature vectors of factorization models. By doing so, one may transfer the well-known accuracy of factorization models observed at the personalized recommendations to the item-to-item case. This paper introduces context-awareness to item similarities in the factorization framework. Two levels of context-aware similarities are defined and applied to two context-aware implicit feedback based factorization methods (iTALS and iTALSx). We investigate the advantages and drawbacks of the approaches on four real life implicit feedback data sets and we characterize the conditions for their application. The results suggest that it is worth using contextual information for item-to-item recommendations in the factorization framework, however, one should carefully select the appropriate method to achieve similar accuracy gain than in the case of the more general item-to-user recommendation scenario. ]]> Thu, 14 Feb 2013 03:24:39 GMT /slideshow/contextaware-similarities-within-the-factorization-framework/16523070 domonkostikk@slideshare.net(domonkostikk) Context-aware similarities within the factorization framework - presented at CARR 2013 domonkostikk Talk deals with a practical recommendation sceaario - item-2-item recommendation (similar/related items) with implicit feedback and context. Solution is provided in the factorization framework. Paper abstract (to appear in ACM digital Library) Item-to-item recommendation - when the most similar items sought to the actual item - is an important recommendation scenario in practical recommender systems. One way to solve this task is to use the similarity between item feature vectors of factorization models. By doing so, one may transfer the well-known accuracy of factorization models observed at the personalized recommendations to the item-to-item case. This paper introduces context-awareness to item similarities in the factorization framework. Two levels of context-aware similarities are defined and applied to two context-aware implicit feedback based factorization methods (iTALS and iTALSx). We investigate the advantages and drawbacks of the approaches on four real life implicit feedback data sets and we characterize the conditions for their application. The results suggest that it is worth using contextual information for item-to-item recommendations in the factorization framework, however, one should carefully select the appropriate method to achieve similar accuracy gain than in the case of the more general item-to-user recommendation scenario. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/casimcarr13-130214032439-phpapp02-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Talk deals with a practical recommendation sceaario - item-2-item recommendation (similar/related items) with implicit feedback and context. Solution is provided in the factorization framework. Paper abstract (to appear in ACM digital Library) Item-to-item recommendation - when the most similar items sought to the actual item - is an important recommendation scenario in practical recommender systems. One way to solve this task is to use the similarity between item feature vectors of factorization models. By doing so, one may transfer the well-known accuracy of factorization models observed at the personalized recommendations to the item-to-item case. This paper introduces context-awareness to item similarities in the factorization framework. Two levels of context-aware similarities are defined and applied to two context-aware implicit feedback based factorization methods (iTALS and iTALSx). We investigate the advantages and drawbacks of the approaches on four real life implicit feedback data sets and we characterize the conditions for their application. The results suggest that it is worth using contextual information for item-to-item recommendations in the factorization framework, however, one should carefully select the appropriate method to achieve similar accuracy gain than in the case of the more general item-to-user recommendation scenario.
Context-aware similarities within the factorization framework - presented at CARR 2013 from Domonkos Tikk
]]> 529 2 https://cdn.slidesharecdn.com/ss_thumbnails/casimcarr13-130214032439-phpapp02-thumbnail.jpg?width=120&height=120&fit=bounds presentation White http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 ݺߣs from CARR 2012 WS - Enhancing Matrix Factorization Through Initialization for Implicit Feedback Databases /slideshow/mf-init-carr12/14884617 mfinitcarr12-121025104615-phpapp01
Executive summary: The paper propose a new method to solve the cold start problem for matrix factorization using metadata. The method works with the realistic implicit feedback scenario. With a smart initialization of the feature matrices better performance values were achieved on several data sets. Paper abstract: The implicit feedback based recommendation problem— when only the user history is available but there are no ratings—is a much harder task than the explicit feedback based recommendation problem, due to the inherent uncertainty of the interpretation of such user feedbacks. Still, this practically important recommendation task received less attention and therefore there are only a few common implicit feedback based algorithms and benchmark datasets. This paper focuses on a common matrix factorization method for the implicit problem and investigates if recommendation performance can be improved by appropriate initialization of the feature vectors before training. We present a general initialization framework that preserves the similarity between entities (users/items) when creating the initial feature vectors, where similarity is defined using e.g. context or metadata information. We demonstrate how the proposed initialization framework can be coupled with MF algorithms. The efficiency of the initialization is evaluated using various context and metadata based similarity concepts on two implicit variants of the MovieLens 10M dataset and one real life implicit database. It is shown that performance gain can attain 10% improvement in recall@50 and in AUC@50.]]>
Executive summary: The paper propose a new method to solve the cold start problem for matrix factorization using metadata. The method works with the realistic implicit feedback scenario. With a smart initialization of the feature matrices better performance values were achieved on several data sets. Paper abstract: The implicit feedback based recommendation problem— when only the user history is available but there are no ratings—is a much harder task than the explicit feedback based recommendation problem, due to the inherent uncertainty of the interpretation of such user feedbacks. Still, this practically important recommendation task received less attention and therefore there are only a few common implicit feedback based algorithms and benchmark datasets. This paper focuses on a common matrix factorization method for the implicit problem and investigates if recommendation performance can be improved by appropriate initialization of the feature vectors before training. We present a general initialization framework that preserves the similarity between entities (users/items) when creating the initial feature vectors, where similarity is defined using e.g. context or metadata information. We demonstrate how the proposed initialization framework can be coupled with MF algorithms. The efficiency of the initialization is evaluated using various context and metadata based similarity concepts on two implicit variants of the MovieLens 10M dataset and one real life implicit database. It is shown that performance gain can attain 10% improvement in recall@50 and in AUC@50.]]> Thu, 25 Oct 2012 10:46:13 GMT /slideshow/mf-init-carr12/14884617 domonkostikk@slideshare.net(domonkostikk) ݺߣs from CARR 2012 WS - Enhancing Matrix Factorization Through Initialization for Implicit Feedback Databases domonkostikk Executive summary: The paper propose a new method to solve the cold start problem for matrix factorization using metadata. The method works with the realistic implicit feedback scenario. With a smart initialization of the feature matrices better performance values were achieved on several data sets. Paper abstract: The implicit feedback based recommendation problem— when only the user history is available but there are no ratings—is a much harder task than the explicit feedback based recommendation problem, due to the inherent uncertainty of the interpretation of such user feedbacks. Still, this practically important recommendation task received less attention and therefore there are only a few common implicit feedback based algorithms and benchmark datasets. This paper focuses on a common matrix factorization method for the implicit problem and investigates if recommendation performance can be improved by appropriate initialization of the feature vectors before training. We present a general initialization framework that preserves the similarity between entities (users/items) when creating the initial feature vectors, where similarity is defined using e.g. context or metadata information. We demonstrate how the proposed initialization framework can be coupled with MF algorithms. The efficiency of the initialization is evaluated using various context and metadata based similarity concepts on two implicit variants of the MovieLens 10M dataset and one real life implicit database. It is shown that performance gain can attain 10% improvement in recall@50 and in AUC@50. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/mfinitcarr12-121025104615-phpapp01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Executive summary: The paper propose a new method to solve the cold start problem for matrix factorization using metadata. The method works with the realistic implicit feedback scenario. With a smart initialization of the feature matrices better performance values were achieved on several data sets. Paper abstract: The implicit feedback based recommendation problem— when only the user history is available but there are no ratings—is a much harder task than the explicit feedback based recommendation problem, due to the inherent uncertainty of the interpretation of such user feedbacks. Still, this practically important recommendation task received less attention and therefore there are only a few common implicit feedback based algorithms and benchmark datasets. This paper focuses on a common matrix factorization method for the implicit problem and investigates if recommendation performance can be improved by appropriate initialization of the feature vectors before training. We present a general initialization framework that preserves the similarity between entities (users/items) when creating the initial feature vectors, where similarity is defined using e.g. context or metadata information. We demonstrate how the proposed initialization framework can be coupled with MF algorithms. The efficiency of the initialization is evaluated using various context and metadata based similarity concepts on two implicit variants of the MovieLens 10M dataset and one real life implicit database. It is shown that performance gain can attain 10% improvement in recall@50 and in AUC@50.
ݺߣs from CARR 2012 WS - Enhancing Matrix Factorization Through Initialization for Implicit Feedback Databases from Domonkos Tikk
]]> 910 2 https://cdn.slidesharecdn.com/ss_thumbnails/mfinitcarr12-121025104615-phpapp01-thumbnail.jpg?width=120&height=120&fit=bounds presentation White http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Fast ALS-Based Tensor Factorization for Context-Aware Recommendation from Implicit Feedback - slides of our ECML PKDD paper /slideshow/itals-ecml12/14884514 italsecml12-121025104007-phpapp01
Executive summary: The paper deals with the implicit feedback problem for recommender systems, and propose a fast context-aware tensor factorization method that can integrate any kind of contextual information. Paper abstract: Albeit the implicit feedback based recommendation problem—when only the user history is available but there are no ratings—is the most typical setting in real-world applications, it is much less researched than the explicit feedback case. State-of-the-art algorithms that are efficient on the explicit case cannot be straightforwardly transformed to the implicit case if scalability should be maintained. There are few implicit feedback benchmark datasets, therefore new ideas are usually experimented on explicit benchmarks. In this paper, we pro- pose a generic context-aware implicit feedback recommender algorithm, coined iTALS. iTALSapplies a fast, ALS-basedtensor factorization learn- ing method that scales linearly with the number of non-zero elements in the tensor. The method also allows us to incorporate various contex- tual information into the model while maintaining its computational effi- ciency. We present two context-aware implementation variants of iTALS. The first incorporates seasonality and enables to distinguish user behav- ior in different time intervals. The other views the user history as sequen- tial information and has the ability to recognize usage pattern typical to certain group of items, e.g. to automatically tell apart product types that are typically purchased repetitively or once. Experiments performed on five implicit datasets (LastFM 1K, Grocery, VoD, and “implicitized” Netflix and MovieLens 10M) show that by integrating context-aware information with our factorization framework into the state-of-the-art implicit recommender algorithm the recommendation quality improves significantly.]]>
Executive summary: The paper deals with the implicit feedback problem for recommender systems, and propose a fast context-aware tensor factorization method that can integrate any kind of contextual information. Paper abstract: Albeit the implicit feedback based recommendation problem—when only the user history is available but there are no ratings—is the most typical setting in real-world applications, it is much less researched than the explicit feedback case. State-of-the-art algorithms that are efficient on the explicit case cannot be straightforwardly transformed to the implicit case if scalability should be maintained. There are few implicit feedback benchmark datasets, therefore new ideas are usually experimented on explicit benchmarks. In this paper, we pro- pose a generic context-aware implicit feedback recommender algorithm, coined iTALS. iTALSapplies a fast, ALS-basedtensor factorization learn- ing method that scales linearly with the number of non-zero elements in the tensor. The method also allows us to incorporate various contex- tual information into the model while maintaining its computational effi- ciency. We present two context-aware implementation variants of iTALS. The first incorporates seasonality and enables to distinguish user behav- ior in different time intervals. The other views the user history as sequen- tial information and has the ability to recognize usage pattern typical to certain group of items, e.g. to automatically tell apart product types that are typically purchased repetitively or once. Experiments performed on five implicit datasets (LastFM 1K, Grocery, VoD, and “implicitized” Netflix and MovieLens 10M) show that by integrating context-aware information with our factorization framework into the state-of-the-art implicit recommender algorithm the recommendation quality improves significantly.]]> Thu, 25 Oct 2012 10:40:05 GMT /slideshow/itals-ecml12/14884514 domonkostikk@slideshare.net(domonkostikk) Fast ALS-Based Tensor Factorization for Context-Aware Recommendation from Implicit Feedback - slides of our ECML PKDD paper domonkostikk Executive summary: The paper deals with the implicit feedback problem for recommender systems, and propose a fast context-aware tensor factorization method that can integrate any kind of contextual information. Paper abstract: Albeit the implicit feedback based recommendation problem—when only the user history is available but there are no ratings—is the most typical setting in real-world applications, it is much less researched than the explicit feedback case. State-of-the-art algorithms that are efficient on the explicit case cannot be straightforwardly transformed to the implicit case if scalability should be maintained. There are few implicit feedback benchmark datasets, therefore new ideas are usually experimented on explicit benchmarks. In this paper, we pro- pose a generic context-aware implicit feedback recommender algorithm, coined iTALS. iTALSapplies a fast, ALS-basedtensor factorization learn- ing method that scales linearly with the number of non-zero elements in the tensor. The method also allows us to incorporate various contex- tual information into the model while maintaining its computational effi- ciency. We present two context-aware implementation variants of iTALS. The first incorporates seasonality and enables to distinguish user behav- ior in different time intervals. The other views the user history as sequen- tial information and has the ability to recognize usage pattern typical to certain group of items, e.g. to automatically tell apart product types that are typically purchased repetitively or once. Experiments performed on five implicit datasets (LastFM 1K, Grocery, VoD, and “implicitized” Netflix and MovieLens 10M) show that by integrating context-aware information with our factorization framework into the state-of-the-art implicit recommender algorithm the recommendation quality improves significantly. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/italsecml12-121025104007-phpapp01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Executive summary: The paper deals with the implicit feedback problem for recommender systems, and propose a fast context-aware tensor factorization method that can integrate any kind of contextual information. Paper abstract: Albeit the implicit feedback based recommendation problem—when only the user history is available but there are no ratings—is the most typical setting in real-world applications, it is much less researched than the explicit feedback case. State-of-the-art algorithms that are efficient on the explicit case cannot be straightforwardly transformed to the implicit case if scalability should be maintained. There are few implicit feedback benchmark datasets, therefore new ideas are usually experimented on explicit benchmarks. In this paper, we pro- pose a generic context-aware implicit feedback recommender algorithm, coined iTALS. iTALSapplies a fast, ALS-basedtensor factorization learn- ing method that scales linearly with the number of non-zero elements in the tensor. The method also allows us to incorporate various contex- tual information into the model while maintaining its computational effi- ciency. We present two context-aware implementation variants of iTALS. The first incorporates seasonality and enables to distinguish user behav- ior in different time intervals. The other views the user history as sequen- tial information and has the ability to recognize usage pattern typical to certain group of items, e.g. to automatically tell apart product types that are typically purchased repetitively or once. Experiments performed on five implicit datasets (LastFM 1K, Grocery, VoD, and “implicitized” Netflix and MovieLens 10M) show that by integrating context-aware information with our factorization framework into the state-of-the-art implicit recommender algorithm the recommendation quality improves significantly.
Fast ALS-Based Tensor Factorization for Context-Aware Recommendation from Implicit Feedback - slides of our ECML PKDD paper from Domonkos Tikk
]]> 1451 2 https://cdn.slidesharecdn.com/ss_thumbnails/italsecml12-121025104007-phpapp01-thumbnail.jpg?width=120&height=120&fit=bounds presentation White http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Recommender Systems Evaluation: A 3D Benchmark - presented at RUE 2012 workshop at ACM Recsys 2012 /slideshow/rue12-recommender-systems-evaluation-14334091/14334091 rue12-recommendersystemsevaluation-120918113918-phpapp01
Recommender systems add value to vast content resources by matching users with items of interest. In recent years, immense progress has been made in recommendation techniques. The evaluation of these has however not been matched and is threatening to impede the further development of recommender systems. In this paper we propose an approach that addresses this impasse by formulating a novel evaluation concept adopting aspects from recommender systems research and industry. Our model can express the quality of a recommender algorithm from three perspectives, the end consumer (user), the service provider and the vendor (business and technique for both). We review current benchmarking activities and point out their shortcomings, which are addressed by our model. We also explain how our 3D benchmarking framework would apply to a specific use case.]]>
Recommender systems add value to vast content resources by matching users with items of interest. In recent years, immense progress has been made in recommendation techniques. The evaluation of these has however not been matched and is threatening to impede the further development of recommender systems. In this paper we propose an approach that addresses this impasse by formulating a novel evaluation concept adopting aspects from recommender systems research and industry. Our model can express the quality of a recommender algorithm from three perspectives, the end consumer (user), the service provider and the vendor (business and technique for both). We review current benchmarking activities and point out their shortcomings, which are addressed by our model. We also explain how our 3D benchmarking framework would apply to a specific use case.]]> Tue, 18 Sep 2012 11:39:16 GMT /slideshow/rue12-recommender-systems-evaluation-14334091/14334091 domonkostikk@slideshare.net(domonkostikk) Recommender Systems Evaluation: A 3D Benchmark - presented at RUE 2012 workshop at ACM Recsys 2012 domonkostikk Recommender systems add value to vast content resources by matching users with items of interest. In recent years, immense progress has been made in recommendation techniques. The evaluation of these has however not been matched and is threatening to impede the further development of recommender systems. In this paper we propose an approach that addresses this impasse by formulating a novel evaluation concept adopting aspects from recommender systems research and industry. Our model can express the quality of a recommender algorithm from three perspectives, the end consumer (user), the service provider and the vendor (business and technique for both). We review current benchmarking activities and point out their shortcomings, which are addressed by our model. We also explain how our 3D benchmarking framework would apply to a specific use case. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/rue12-recommendersystemsevaluation-120918113918-phpapp01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Recommender systems add value to vast content resources by matching users with items of interest. In recent years, immense progress has been made in recommendation techniques. The evaluation of these has however not been matched and is threatening to impede the further development of recommender systems. In this paper we propose an approach that addresses this impasse by formulating a novel evaluation concept adopting aspects from recommender systems research and industry. Our model can express the quality of a recommender algorithm from three perspectives, the end consumer (user), the service provider and the vendor (business and technique for both). We review current benchmarking activities and point out their shortcomings, which are addressed by our model. We also explain how our 3D benchmarking framework would apply to a specific use case.
Recommender Systems Evaluation: A 3D Benchmark - presented at RUE 2012 workshop at ACM Recsys 2012 from Domonkos Tikk
]]> 1249 4 https://cdn.slidesharecdn.com/ss_thumbnails/rue12-recommendersystemsevaluation-120918113918-phpapp01-thumbnail.jpg?width=120&height=120&fit=bounds presentation White http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 From a toolkit of recommendation algorithms into a real business: the Gravity R&D experience (talk given at Recsys 2012) /domonkostikk/from-a-toolkit-of-recommendation-algorithms-into-a-real-business-the-gravity-rd-experience recsyschallengews-120914055336-phpapp02
Talk given at Recsys Challenge Workshop in Dublin (@ ACM Recsys 2012), on 13.09.2012.]]>
Talk given at Recsys Challenge Workshop in Dublin (@ ACM Recsys 2012), on 13.09.2012.]]> Fri, 14 Sep 2012 05:53:35 GMT /domonkostikk/from-a-toolkit-of-recommendation-algorithms-into-a-real-business-the-gravity-rd-experience domonkostikk@slideshare.net(domonkostikk) From a toolkit of recommendation algorithms into a real business: the Gravity R&D experience (talk given at Recsys 2012) domonkostikk Talk given at Recsys Challenge Workshop in Dublin (@ ACM Recsys 2012), on 13.09.2012. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/recsyschallengews-120914055336-phpapp02-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Talk given at Recsys Challenge Workshop in Dublin (@ ACM Recsys 2012), on 13.09.2012.
From a toolkit of recommendation algorithms into a real business: the Gravity R&D experience (talk given at Recsys 2012) from Domonkos Tikk
]]> 1675 2 https://cdn.slidesharecdn.com/ss_thumbnails/recsyschallengews-120914055336-phpapp02-thumbnail.jpg?width=120&height=120&fit=bounds presentation White http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 https://cdn.slidesharecdn.com/profile-photo-domonkostikk-48x48.jpg?cb=1688557619 Gravity R&D is a recommendation engine provider, using machine learning to personalize digital customer experiences for SMEs and enterprises. The Budapest-based company has been focusing on data science since 2009, using machine learning and Big Data analytics to create personalized customer experiences for brands in various industries. Gravity's products, Yusp and Yuspify, help clients deliver better brand experiences, drive revenue growth and improve customer satisfaction. The company's personalization solutions easily serves 35+ billion personalized recommendations per month. Gravity is strong in R&D, and proud to have a data mining team active in the field of recommender systems. www.yusp.com https://cdn.slidesharecdn.com/ss_thumbnails/gravitylessonslearntatbuildingrecservicesatindustryscalefinal-160324094257-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/lessons-learnt-at-building-recommendation-services-at-industry-scale/59982420 Lessons learnt at buil... https://cdn.slidesharecdn.com/ss_thumbnails/gravitymedianet2015-151011085335-lva1-app6891-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/recommenders-on-video-sharing-portals-business-and-algorithmic-aspects/53789135 Recommenders on video ... https://cdn.slidesharecdn.com/ss_thumbnails/gravitylsrs2015recsysneighbourmethodsvsmf-150922080707-lva1-app6891-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/neighbor-methods-vs-matrix-factorization-case-studies-of-reallife-recommendations-gravity-lsrs2015-recsys-2015/53051002 Neighbor methods vs ma...