Congratulations! Your paper, “Recommendation Through Mixtures of Heterogeneous Item Relationships” (submission number 1613), has been accepted for inclusion in the CIKM 2018 conference. We received 862 full paper submissions, out of which 147 were accepted (17%). The reviews and metareview of your paper are at the bottom of this email. A separate email will be sent with instructions for preparation of the camera-ready version of the paper. Please take full account of the reviewers' comments when preparing the camera-ready version. CIKM requires that one of the authors of your paper attend the conference and present your work. Please also consider attending tutorials and workshops in the conference (http://www.cikm2018.units.it/). Best regards, James Allan, Norman Paton, Divesh Srivastava CIKM 2018 Long Paper Track Co-Chairs ----------------------- REVIEW 1 --------------------- PAPER: 1613 TITLE: Recommendation Through Mixtures of Heterogeneous Item Relationships AUTHORS: Wang-Cheng Kang, Mengting Wan and Julian Mcauley Relevance to CIKM: 4 (good) Originality of the Work: 4 (good) Technical Soundness: 4 (good) Quality of Presentation: 4 (good) Impact of Ideas or Results: 4 (good) Adequacy of Citations: 4 (good) Reproducibility of Methods: 4 (good) Overall Evaluation: 3 (I half-champion and would accept if someone else is also at least half-championing) ----------- List 3 or more strong points, labelled S1, S2, ... ----------- S1. motivation is clear and practical S2. nicely combined multiple relationships into a one framework S3. Experiment results support all the claims. ----------- List 3 or more weak points, labelled W1, W2, ... ----------- W1. N/A ----------- Overall Evaluation ----------- This paper proposes a new method, MoHR. The proposed method generalizes the mechanism for considering the relationship between items by modeling relationships as latent vectors. In addition, this method combines relationship prediction model and sequential recommendation technique into the model. The motivation of this paper is clear and practical since there are multiple methods considering the different relationship among items. In addition, providing a reason for user’s selection by incorporating the relationship prediction has strong practical usage. All the stages are successfully combined without any particular flaw, and evaluation results also support the claims that the MoHR can deal with multiple relationships and also can provide the explanation of item choice. For these reasons, I vote for a accept. ----------------------- REVIEW 2 --------------------- PAPER: 1613 TITLE: Recommendation Through Mixtures of Heterogeneous Item Relationships AUTHORS: Wang-Cheng Kang, Mengting Wan and Julian Mcauley Relevance to CIKM: 4 (good) Originality of the Work: 4 (good) Technical Soundness: 4 (good) Quality of Presentation: 4 (good) Impact of Ideas or Results: 4 (good) Adequacy of Citations: 4 (good) Reproducibility of Methods: 4 (good) Overall Evaluation: 3 (I half-champion and would accept if someone else is also at least half-championing) ----------- List 3 or more strong points, labelled S1, S2, ... ----------- S1. The paper proposes to consider a mixture of heterogeneous relationship types, which is a novel idea. S2. The paper is well written. The proposed method is technically sound. S3. Experiments are extensive and the results are promising. ----------- List 3 or more weak points, labelled W1, W2, ... ----------- None ----------- Overall Evaluation ----------- The paper explores a novel idea that leverages a mixture of heterogeneous relationships. The paper is very well written, with a technically sound solution, and reasonable complexity / model size. The experiments are extensive, with both quantitative and qualitative results, covering multiple datasets with various competitive baselines, and various analysis such as ablation study and hyper-parameter analysis to support their claims. ----------------------- REVIEW 3 --------------------- PAPER: 1613 TITLE: Recommendation Through Mixtures of Heterogeneous Item Relationships AUTHORS: Wang-Cheng Kang, Mengting Wan and Julian Mcauley Relevance to CIKM: 5 (excellent) Originality of the Work: 4 (good) Technical Soundness: 4 (good) Quality of Presentation: 4 (good) Impact of Ideas or Results: 4 (good) Adequacy of Citations: 5 (excellent) Reproducibility of Methods: 4 (good) Overall Evaluation: 3 (I half-champion and would accept if someone else is also at least half-championing) ----------- List 3 or more strong points, labelled S1, S2, ... ----------- S1. An important problem in recommendation systems with real impact S2. Utilizing heterogeneous item relationships S3. Modeling different aspects of recommendation in a principled way ----------- List 3 or more weak points, labelled W1, W2, ... ----------- W1. It is hard to tuning the parameters in the objective function ----------- Overall Evaluation ----------- This paper tries to utilize the combination of different signals in the recommendation system, including different types of relationships and sequential information. Each of these aspects is modeled as a component in the objective function, and they together are unified and optimized in parallel under a multi-task learning framework. The way to unify different types of relationships and signals is principled and shown to be effective in experiments. My only concern is that the meaning of hyper-parameters in the objectives is not very clear - especially between between beta and lambda. And it is not easy to tune these parameters ------------------------- METAREVIEW ------------------------ PAPER: 1613 TITLE: Recommendation Through Mixtures of Heterogeneous Item Relationships The authors propose a probabilistic mixture model for recommendations based on heterogeneous data sources. The reviewers appreciate the relevance and timeliness of the problem. The proposed solution is clearly motivated and described. It is deemed sound and somewhat novel, although it could be better stated in contrast to related efforts (e.g., RecSys hosted the HetRec workshop on the topic in 2010 and 2011). The conducted evaluation is also sound and insightful, and demonstrates the effectiveness of the proposal in contrast to several state-of-the-art baselines across several datasets.