Jose Ignacio Honrado, Oscar Huarte, Cesar Jimenez, Sebastian Ortega, José R. Pérez-Agüera, Joaquín Pérez-Iglesias, Álvaro Polo, Gabriel Rodríguez
In this paper we describe the system built by the Jobandtalent Recommendation Team to compete in the RecSys Challenge 2016. The task consisted in predicting future interactions between Users and Items within the XING platform. The data provided by XING consists of users, items, plus interactions, and impressions of items showed to those users. We decided to apply a Learning to Rank approach to find the best combination of relevance features. We finally achieved the 11th position.
{"title":"Jobandtalent at RecSys Challenge 2016","authors":"Jose Ignacio Honrado, Oscar Huarte, Cesar Jimenez, Sebastian Ortega, José R. Pérez-Agüera, Joaquín Pérez-Iglesias, Álvaro Polo, Gabriel Rodríguez","doi":"10.1145/2987538.2987547","DOIUrl":"https://doi.org/10.1145/2987538.2987547","url":null,"abstract":"In this paper we describe the system built by the Jobandtalent Recommendation Team to compete in the RecSys Challenge 2016. The task consisted in predicting future interactions between Users and Items within the XING platform. The data provided by XING consists of users, items, plus interactions, and impressions of items showed to those users. We decided to apply a Learning to Rank approach to find the best combination of relevance features. We finally achieved the 11th position.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131806453","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper describes our solution for the RecSys Challenge 2016. In the challenge, several datasets were provided from a social network for business XING. The goal of the competition was to use these data to predict job postings that a user will interact positively with (click, bookmark or reply). Our solution to this problem includes three different types of models: Factorization Machine, item-based collaborative filtering, and content-based topic model on tags. Thus, we combined collaborative and content-based approaches in our solution. Our best submission, which was a blend of ten models, achieved 7th place in the challenge's final leader-board with a score of 1677 898.52. The approaches presented in this paper are general and scalable. Therefore they can be applied to another problem of this type.
{"title":"Job recommendation based on factorization machine and topic modelling","authors":"V. Leksin, A. Ostapets","doi":"10.1145/2987538.2987542","DOIUrl":"https://doi.org/10.1145/2987538.2987542","url":null,"abstract":"This paper describes our solution for the RecSys Challenge 2016. In the challenge, several datasets were provided from a social network for business XING. The goal of the competition was to use these data to predict job postings that a user will interact positively with (click, bookmark or reply). Our solution to this problem includes three different types of models: Factorization Machine, item-based collaborative filtering, and content-based topic model on tags. Thus, we combined collaborative and content-based approaches in our solution. Our best submission, which was a blend of ten models, achieved 7th place in the challenge's final leader-board with a score of 1677 898.52. The approaches presented in this paper are general and scalable. Therefore they can be applied to another problem of this type.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129752991","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recommendation Systems are omnipresent on the web nowadays. Most websites today are striving to provide quality recommendations to their customers in order to increase and retain their customers. In this paper, we present our approaches to design a job recommendation system for a career based social networking website - XING. We take a bottom up approach: we start with deeply understanding and exploring the data and gradually build the smaller bits of the system. We also consider traditional approaches of recommendation systems like collaborative filtering and discuss its performance. The best model that we produced is based on Gradient Boosting algorithm. Our experiments show the efficacy of our approaches. This work is based on a challenge organized by ACM RecSys conference 2016. We achieved a final full score of 1,411,119.11 with rank 20 on the official leader board.
{"title":"A bottom-up approach to job recommendation system","authors":"Sonu K. Mishra, Manoj Reddy","doi":"10.1145/2987538.2987546","DOIUrl":"https://doi.org/10.1145/2987538.2987546","url":null,"abstract":"Recommendation Systems are omnipresent on the web nowadays. Most websites today are striving to provide quality recommendations to their customers in order to increase and retain their customers. In this paper, we present our approaches to design a job recommendation system for a career based social networking website - XING. We take a bottom up approach: we start with deeply understanding and exploring the data and gradually build the smaller bits of the system. We also consider traditional approaches of recommendation systems like collaborative filtering and discuss its performance. The best model that we produced is based on Gradient Boosting algorithm. Our experiments show the efficacy of our approaches. This work is based on a challenge organized by ACM RecSys conference 2016. We achieved a final full score of 1,411,119.11 with rank 20 on the official leader board.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127757877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The present paper introduces a solution for the RecSys Challenge 2016. The principle of the proposed technique is to define various models capturing the specificity of the dataset and then to subsequently find the optimal combinations of these by considering different user categories. The approach follows a practical way for the fine-tuning of recommender algorithms, highlighting their components, training-and prediction time. Based on forward predictor selection, it can be shown that item-neighbor methods and the recommendation of already shown or interacted items have great potential in improving the offline accuracy. The best composition consists of 11 predictor instances that achieved the third place with 665,592 leaderboard score and 2,005,263 final score.
{"title":"A combination of simple models by forward predictor selection for job recommendation","authors":"Dávid Zibriczky","doi":"10.1145/2987538.2987548","DOIUrl":"https://doi.org/10.1145/2987538.2987548","url":null,"abstract":"The present paper introduces a solution for the RecSys Challenge 2016. The principle of the proposed technique is to define various models capturing the specificity of the dataset and then to subsequently find the optimal combinations of these by considering different user categories. The approach follows a practical way for the fine-tuning of recommender algorithms, highlighting their components, training-and prediction time. Based on forward predictor selection, it can be shown that item-neighbor methods and the recommendation of already shown or interacted items have great potential in improving the offline accuracy. The best composition consists of 11 predictor instances that achieved the third place with 665,592 leaderboard score and 2,005,263 final score.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117080194","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, we present an ensemble method for job recommendation to ACM RecSys Challenge 2016. Given a user, the goal of a job recommendation system is to predict those job postings that are likely to be relevant to the user1. Firstly, we analyze the train dataset and find several interesting patterns. Secondly, we describe our solution, which is an ensemble of two filters, combining the merits of traditional collaborative filtering and content-based filtering. Our approach finally achieved a score of 1632828.82, ranked at the 10th place on the public leaderboard.
{"title":"An ensemble method for job recommender systems","authors":"Chenrui Zhang, Xueqi Cheng","doi":"10.1145/2987538.2987545","DOIUrl":"https://doi.org/10.1145/2987538.2987545","url":null,"abstract":"In this paper, we present an ensemble method for job recommendation to ACM RecSys Challenge 2016. Given a user, the goal of a job recommendation system is to predict those job postings that are likely to be relevant to the user1.\u0000 Firstly, we analyze the train dataset and find several interesting patterns. Secondly, we describe our solution, which is an ensemble of two filters, combining the merits of traditional collaborative filtering and content-based filtering. Our approach finally achieved a score of 1632828.82, ranked at the 10th place on the public leaderboard.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124590717","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper we present our method used in the RecSys '16 Challenge. In particular, we propose a general collaborative filtering framework where many predictors can be cast. The framework is able to incorporate information about the content but in a collaborative fashion. Using this framework we instantiate a set of different predictors that consider different aspects of the dataset provided for the challenge. In order to merge all these aspects together, we also provide a method able to linearly combine the predictors. This method learns the weights of the predictors by solving a quadratic optimization problem. In the experimental section we show the performance using different predictors combinations. Results highlight the fact that the combination always outperforms the single predictor.
{"title":"A preliminary study on a recommender system for the job recommendation challenge","authors":"Mirko Polato, F. Aiolli","doi":"10.1145/2987538.2987549","DOIUrl":"https://doi.org/10.1145/2987538.2987549","url":null,"abstract":"In this paper we present our method used in the RecSys '16 Challenge.\u0000 In particular, we propose a general collaborative filtering framework where many predictors can be cast. The framework is able to incorporate information about the content but in a collaborative fashion. Using this framework we instantiate a set of different predictors that consider different aspects of the dataset provided for the challenge. In order to merge all these aspects together, we also provide a method able to linearly combine the predictors. This method learns the weights of the predictors by solving a quadratic optimization problem.\u0000 In the experimental section we show the performance using different predictors combinations. Results highlight the fact that the combination always outperforms the single predictor.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126944827","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recommender systems can be used as a tool to assist people in finding a job. However, this specific domain requires expert algorithms with domain knowledge to recommend jobs conformable to people's expertise and interests. This is the topic of the Recsys Challenge 2016, which aims for an algorithm that predicts the job postings that a user will positively interact with. Our solution is a hybrid algorithm combining a content-based and KNN approach. The content-based algorithm matches features of candidate recommendations and job postings of historical interactions. The KNN approach searches for the job postings that are the most similar to the postings the user interacted with in the past. The resulting combination is a lightweight algorithm that is fast and scalable, generating recommendations with a proper evaluation score.
{"title":"A scalable, high-performance Algorithm for hybrid job recommendations","authors":"Toon De Pessemier, K. Vanhecke, L. Martens","doi":"10.1145/2987538.2987539","DOIUrl":"https://doi.org/10.1145/2987538.2987539","url":null,"abstract":"Recommender systems can be used as a tool to assist people in finding a job. However, this specific domain requires expert algorithms with domain knowledge to recommend jobs conformable to people's expertise and interests. This is the topic of the Recsys Challenge 2016, which aims for an algorithm that predicts the job postings that a user will positively interact with. Our solution is a hybrid algorithm combining a content-based and KNN approach. The content-based algorithm matches features of candidate recommendations and job postings of historical interactions. The KNN approach searches for the job postings that are the most similar to the postings the user interacted with in the past. The resulting combination is a lightweight algorithm that is fast and scalable, generating recommendations with a proper evaluation score.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115283642","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
T. Carpi, Marco Edemanti, Ervin Kamberoski, Elena Sacchi, P. Cremonesi, Roberto Pagano, Massimo Quadrana
This paper describes the approach that team PumpkinPie adopted in the 2016 Recsys Challenge. The task of the competition organized by XING is to predict which job postings the user has interacted with. The team's approach mainly consists in generating a set of models using different techniques, and then combining them in a multi-stack ensemble. This strategy granted the fourth position in the final leader-board to the team, with an overall score of 1.86M.
{"title":"Multi-stack ensemble for job recommendation","authors":"T. Carpi, Marco Edemanti, Ervin Kamberoski, Elena Sacchi, P. Cremonesi, Roberto Pagano, Massimo Quadrana","doi":"10.1145/2987538.2987541","DOIUrl":"https://doi.org/10.1145/2987538.2987541","url":null,"abstract":"This paper describes the approach that team PumpkinPie adopted in the 2016 Recsys Challenge. The task of the competition organized by XING is to predict which job postings the user has interacted with. The team's approach mainly consists in generating a set of models using different techniques, and then combining them in a multi-stack ensemble. This strategy granted the fourth position in the final leader-board to the team, with an overall score of 1.86M.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130388485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Wen-Li Xiao, Xiao Xu, Kang Liang, Junkang Mao, Jun Wang
The RecSys Challenge 2016 focuses on the prediction of users' interest in clicking a job posting in the career-oriented social networking site Xing. Given users' profile, the content of the job posting, as well as the historical activities of users, we aim in recommending top job postings to users for the coming week. This paper introduces the winning strategy for such a recommendation task. We summarize several key components that result in our leading position in this contest. First, we build a hierarchical pairwise model with ensemble learning as the overall prediction framework. Second, we integrate both content and behavior information in our feature engineering process. In particular, we model the temporal activity pattern using a self-exciting point process, namely Hawkes Process, to generate the most relevant recommendation at the right moment. Finally, we also tackle the challenging cold start issue using a semantic based strategy that is built on the topic modeling with the users profiling information. Our approach achieved the highest leader-board and full scores among all the submissions.
{"title":"Job recommendation with Hawkes process: an effective solution for RecSys Challenge 2016","authors":"Wen-Li Xiao, Xiao Xu, Kang Liang, Junkang Mao, Jun Wang","doi":"10.1145/2987538.2987543","DOIUrl":"https://doi.org/10.1145/2987538.2987543","url":null,"abstract":"The RecSys Challenge 2016 focuses on the prediction of users' interest in clicking a job posting in the career-oriented social networking site Xing. Given users' profile, the content of the job posting, as well as the historical activities of users, we aim in recommending top job postings to users for the coming week. This paper introduces the winning strategy for such a recommendation task. We summarize several key components that result in our leading position in this contest. First, we build a hierarchical pairwise model with ensemble learning as the overall prediction framework. Second, we integrate both content and behavior information in our feature engineering process. In particular, we model the temporal activity pattern using a self-exciting point process, namely Hawkes Process, to generate the most relevant recommendation at the right moment. Finally, we also tackle the challenging cold start issue using a semantic based strategy that is built on the topic modeling with the users profiling information. Our approach achieved the highest leader-board and full scores among all the submissions.","PeriodicalId":127880,"journal":{"name":"RecSys Challenge '16","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126697102","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：