Session-based item recommendation in e-commerce: on short-term intents, reminders, trends and discounts


Many e-commerce sites present additional item recommendations to their visitors while they navigate the site, and ample evidence exists that such recommendations are valuable for both customers and providers. Academic research often focuses on the capability of recommender systems to help users discover items they presumably do not know yet and which match their long-term preference profiles. In reality, however, recommendations can be helpful for customers also for other reasons, for example, when they remind them of items they were recently interested in or when they point site visitors to items that are currently discounted. In this work, we first adopt a systematic statistical approach to analyze what makes recommendations effective in practice and then propose ways of operationalizing these insights into novel recommendation algorithms. Our data analysis is based on log data of a large e-commerce site. It shows that various factors should be considered in parallel when selecting items for recommendation, including their match with the customer’s shopping interests in the previous sessions, the general popularity of the items in the last few days, as well as information about discounts. Based on these analyses we propose a novel algorithm that combines a neighborhood-based scheme with a deep neural network to predict the relevance of items for a given shopping session.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5


  1. 1.

  2. 2.

    The dataset provided to us was fully anonymized and artificially distorted so that no inference on true sales numbers on the site can be made.

  3. 3.

    Details about the used recommendation algorithm on the website were not disclosed to us.

  4. 4.

    Using much larger samples makes some of the experiments reported in later sections computationally challenging as some of the algorithms require extensive hyperparameter tuning.

  5. 5.

    The conversion rate cannot directly be interpreted as the absolute amount of additional sales that is generated by a recommender since we cannot know if an item would have been bought by a user even when it was not recommended. Previous studies however indicate that recommenders in general can turn more visitors into buyers and be effective in terms of generating additional sales (Jannach and Hegelich 2009).

  6. 6.

    Since brand loyalty is common in the fashion domain, a strong effect was generally expected.

  7. 7.

    The popularity measurement only considered view and purchases related to the items up to the time point of the recommendation on that day.

  8. 8.

    See, e.g., Manning et al. (2008) for details about these feature selection methods.

  9. 9.

    The results of the analysis for the set of 3000 occasional users are similar and are reported in “Appendix B”. The detailed results of the feature analysis are provided at

  10. 10.

    The entry “Distance to first/last view” in the table refers to the time that has passed since the user has viewed a recommended item the first or last time before the purchase.

  11. 11.

  12. 12.

    According to a four field Chi-squared test (\(p<0.05\)) with Bonferroni correction.

  13. 13.

    More information about the technical details of the algorithms are provided in Lerche et al. (2016).

  14. 14.

    We unfortunately were not yet able to benchmark C-KNN in a field study with our external partner.

  15. 15.

    The competition was organized in the context of IJCAI ’15 and TMall is a leading Chinese online market place in the style of Amazon (

  16. 16.

    For the sake of brevity we do not report the detailed results for the other datasets here. The additional results, which are in line with the observations for the Zalando dataset, can be found in Lerche et al. (2016).

  17. 17.

    In our empirical evaluation presented in Sect. 4.5 the baseline algorithms were used to create a set of 200 recommendations to be re-ranked, which on average led to the best results in terms of the hit rate.

  18. 18.

    In the experiments reported below we used the frequent user sample, as described in Sect. 2.1, as a basis for learning.

  19. 19.

  20. 20.

  21. 21.

    The features that were not considered by DeepPredict are listed in the appendix in Table 16. All non-considered features are related to the recent popularity of a certain brand or category.

  22. 22.

    y is the correct label (1 or 0) and p the predicted probability.

  23. 23.

    As in our previous experiments, we took the first 200 recommendations of the baseline algorithms as a basis for re-ranking as this cut-off value on average led to the best results in terms of the hit rate.

  24. 24.

    Revealing fewer view items of the current session leads to accuracy values that are lower on an absolute scale; the ranking of the algorithms is however not changed.

  25. 25.

    All reported differences are statistically significant according to a Student’s t-test at \(p<0.01\) with Bonferroni correction.


  1. Adomavicius, G., Tuzhilin, A.: Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions. IEEE Trans. Knowl. Data Eng. 17(6), 734–749 (2005)

    Article  Google Scholar 

  2. Adomavicius, G., Tuzhilin, A.: Context-aware recommender systems. In: Recommender Systems Handbook. Springer, pp. 217–253 (2011)

  3. Aghabozorgi, S.R., Wah, T.Y.: Recommender systems: incremental clustering on web log data. In: Proceedings of the 2nd International Conference on Interaction Sciences: Information Technology, Culture and Human, ICIS ’09, pp. 812–818 (2009)

  4. AlMurtadha, Y., Sulaiman, N.B., Mustapha, N., Udzir, N.I., Muda, Z.: ARS: web page recommendation system for anonymous users based on web usage mining. In: Proceedings of the European Conference of Systems, and European Conference of Circuits Technology and Devices, and European Conference of Communications, and European Conference on Computer Science, ECS’10/ECCTD’10/ECCOM’10/ECCS’10, pp. 115–120 (2010)

  5. Anand, S., Mobasher, B.: Contextual recommendation. In: From Web to Social Web. Springer, pp. 142–160 (2007)

  6. Anderson, A., Kumar, R., Tomkins, A., Vassilvitskii, S.: The dynamics of repeat consumption. In: Proceedings of the 23rd International Conference on World Wide Web, WWW’14, pp. 419–430 (2014)

  7. Berry, S., Levinsohn, J., Pakes, A.: Automobile prices in market equilibrium. Econometrica 63(4), 841–890 (1995)

    Article  MATH  Google Scholar 

  8. Bonnin, G., Jannach, D.: Automated generation of music playlists: survey and experiments. ACM Comput. Surv. 47(2), 26:1–26:35 (2014)

    Article  Google Scholar 

  9. Burke, R.: Knowledge-based recommender systems. Encyclop. Libr. Inf. Sci. 69(32), 180–200 (2000)

    Google Scholar 

  10. Candel, A., Parmar, V., LeDell, E., Arora, A.: Deep learning with \(\text{H}_{2}\text{ O }\). Accessed 17 August 2017

  11. Chen, S., Moore, J.L., Turnbull, D., Joachims, T.: Playlist prediction via metric embedding. In: Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD’12, pp. 714–722 (2012)

  12. Choudhary, V., Ghose, A., Mukhopadhyay, T., Rajan, U.: Personalized pricing and quality differentiation. Manag. Sci. 51(7), 1120–1130 (2005)

    Article  Google Scholar 

  13. Close, A.G., Kukar-Kinney, M.: Beyond buying: motivations behind consumers’ online shopping cart use. J Bus. Res. Adv. Internet Consum. Behav. Mark. Strategy 63(9–10), 986–992 (2010)

    Google Scholar 

  14. Diehl, K., van Herpen, E., Lamberton, C.: Organizing products with complements versus substitutes: effects on store preferences as a function of effort and assortment perceptions. J. Retail. 91(1), 1–18 (2015)

    Article  Google Scholar 

  15. Garcin, F., Dimitrakakis, C., Faltings, B.: Personalized news recommendation with context trees. In: Proceedings of the 7th ACM Conference on Recommender Systems, RecSys’13, pp. 105–112 (2013)

  16. Garcin, F., Faltings, B., Donatsch, O., Alazzawi, A., Bruttin, C., Huber, A.: Offline and online evaluation of news recommender systems at In: Proceedings of the 8th ACM Conference on Recommender Systems, RecSys’14, pp. 169–176 (2014)

  17. Gomez-Uribe, C.A., Hunt, N.: The netflix recommender system: algorithms, business value, and innovation. ACM Trans. Manag. Inf. Syst. 6(4), 13:1–13:19 (2015)

    Article  Google Scholar 

  18. Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. MIT Press. Accessed 17 August 2017 (2016)

  19. Hariri, N., Mobasher, B., Burke, R.: Context-aware music recommendation based on latent topic sequential patterns. In: Proceedings of the Sixth ACM Conference on Recommender Systems, RecSys’12, pp. 131–138 (2012)

  20. Hariri, N., Mobasher, B., Burke, R.: Context adaptation in interactive recommender systems. In: Proceedings of the 8th ACM Conference on Recommender Systems, RecSys’14, pp. 41–48 (2014)

  21. Hidasi, B., Karatzoglou, A., Baltrunas, L., Tikk, D.: Session-based recommendations with recurrent neural networks. In: Proceedings of the International Conference on Learning Representations, ICLR’16 (2016)

  22. Hinton, G.E., Srivastava, N., Krizhevsky, A., Sutskever, I., Salakhutdinov, R.: Improving neural networks by preventing co-adaptation of feature detectors. CoRR arXiv:1207.0580 (2012)

  23. Jannach, D., Adomavicius, G.: Recommendations with a purpose. In: Proceedings of the 10th ACM Conference on Recommender Systems, RecSys’16, pp. 7–10 (2016)

  24. Jannach, D., Hegelich, K.: A case study on the effectiveness of recommendations in the mobile internet. In: Proceedings of the 3rd ACM Conference on Recommender Systems, RecSys’09, pp. 205–208 (2009)

  25. Jannach, D., Ludewig, M.: Determining characteristics of successful recommendations from log data—a case study. In: Proceedings of the Symposium on Applied Computing, SAC’17, pp. 1643–1648 (2017a)

  26. Jannach, D., Ludewig, M.: When recurrent neural networks meet the neighborhood for session-based recommendation. In: Proceedings of the 11th ACM Conference on Recommender Systems, RecSys’17, pp. 306–310 (2017b)

  27. Jannach, D., Lerche, L., Jugovac, M.: Adaptation and evaluation of recommendations for short-term shopping goals. In: Proceedings of the 9th ACM Conference on Recommender Systems, RecSys’15, pp. 211–218 (2015a)

  28. Jannach, D., Lerche, L., Kamehkhosch, I.: Beyond “hitting the hits”—generating coherent music playlist continuations with the right tracks. In: Proceedings of the 9th ACM Conference on Recommender Systems, RecSys’15, pp. 187–194 (2015b)

  29. Jannach, D., Lerche, L., Kamehkhosh, I., Jugovac, M.: What recommenders recommend: an analysis of recommendation biases and possible countermeasures. User Model. User Adapt Interact. 25(5), 427–491 (2015c)

    Article  Google Scholar 

  30. Kamishima, T., Akaho, S.: Personalized pricing recommender system: multi-stage epsilon-greedy approach. In: Proceedings of the 2nd International Workshop on Information Heterogeneity and Fusion in Recommender Systems, HetRec’11, pp. 57–64 (2011)

  31. Kapoor, K., Kumar, V., Terveen, L., Konstan, J.A., Schrater, P.: I like to explore sometimes: adapting to dynamic user novelty preferences. In: Proceedings of the 9th ACM Conference on Recommender Systems, RecSys’15, pp. 19–26 (2015)

  32. Lerche, L., Jannach, D., Ludewig, M.: On the value of reminders within e-commerce recommendations. In: Proceedings of the 2016 Conference on User Modeling Adaptation and Personalization, UMAP’16, pp. 27–25 (2016)

  33. Li, L., Chu, W., Langford, J., Schapire, R.E.: A contextual-bandit approach to personalized news article recommendation. In: Proceedings of the 19th International Conference on World Wide Web, WWW’10, pp. 661–670 (2010)

  34. Liu, J., Dolan, P., Pedersen, E.R.: Personalized news recommendation based on click behavior. In: Proceedings of the 15th International Conference on Intelligent User Interfaces, IUI’10, pp. 31–40 (2010)

  35. Manning, C.D., Raghavan, P., Schütze, H.: Introduction to Information Retrieval. Cambridge University Press, New York, NY (2008)

    Google Scholar 

  36. Mobasher, B., Dai, H., Luo, T., Nakagawa, M.: Using sequential and non-sequential patterns in predictive web usage mining tasks. In: Proceedings of the 2002 IEEE International Conference on Data Mining, ICDM’02, pp. 669–672 (2002)

  37. Moe, W.W.: Buying, searching, or browsing: differentiating between online shoppers using in-store navigational clickstream. J. Consum. Psychol. 13(1), 29–39 (2003)

    Article  Google Scholar 

  38. Nguyen, Q.N., Ricci, F.: Long-term and session-specific user preferences in a mobile recommender system. In: Proceedings of the 13th International Conference on Intelligent User Interfaces, IUI’08, pp. 381–384 (2008)

  39. Padmanabhan, P., Sadekar, K., Krishnan, G.: What’s trending on netflix?. Accessed 17 August 2017 (2015)

  40. Plate, C., Basselin, N., Kröner, A., Schneider, M., Baldes, S., Dimitrova, V., Jameson, A.: Recomindation: New functions for augmented memories. In: Proceedings of the 4th International Conference on Adaptive Hypermedia and Adaptive Web-Based Systems, AH’06, pp. 141–150 (2006)

  41. Prassas, G., Pramataris, K.C., Papaemmanouil, O., Doukidis, G.J.: A recommender system for online shopping based on past customer behaviour. In: Proceedings of the 14th BLED Electronic Commerce Conference, BLED’01, pp. 766–782 (2001)

  42. Rendle, S., Freudenthaler, C., Gantner, Z., Schmidt-Thieme, L. (2009) BPR: Bayesian personalized ranking from implicit feedback. In: Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence, UAI’09, pp. 452–461

  43. Rendle, S., Freudenthaler, C., Schmidt-Thieme, L.: Factorizing personalized markov chains for next-basket recommendation. In: Proceedings of the 19th International Conference on World Wide Web, WWW’10, pp. 811–820 (2010)

  44. Ricci, F., Venturini, A., Cavada, D., Mirzadeh, N., Blaas, D., Nones, M.: Product recommendation with interactive query management and twofold similarity. In: Proceedings of the 5th International Conference on Case-Based Reasoning, ICCBR’03, pp. 479–493 (2003)

  45. Romov, P., Sokolov, E.: Recsys challenge 2015: ensemble learning with categorical features. In: Proceedings of the 2015 International ACM Recommender Systems Challenge, RecSys’15 Challenge, pp. 1:1–1:4 (2015)

  46. Schnabel, T., Bennett, P.N., Dumais, S.T., Joachims, T.: Using shortlists to support decision making and improve recommender system performance. In: Proceedings of the 25th International Conference on World Wide Web, WWW’16, pp. 987–997 (2016)

  47. Shani, G., Heckerman, D., Brafman, R.I.: An MDP-based recommender system. J. Mach. Learn. Res. 6, 1265–1295 (2005)

    MATH  MathSciNet  Google Scholar 

  48. Shen, E., Lieberman, H., Lam, F.: What am I gonna wear?: scenario-oriented recommendation. In: Proceedings of the 12th International Conference on Intelligent User Interfaces, IUI’07, pp. 365–368 (2007)

  49. Sutskever, I., Martens, J., Dahl, G., Hinton, G.: On the importance of initialization and momentum in deep learning. In: Proceedings of the 30th International Conference on International Conference on Machine Learning, ICML’13, pp. 1139–1147 (2013)

  50. Tavakol, M., Brefeld, U.: Factored MDPs for detecting topics of user sessions. In: Proceedings of the 8th ACM Conference on Recommender Systems, RecSys’14, pp. 33–40 (2014)

  51. Wager, S., Wang, S., Liang, P.: Dropout training as adaptive regularization. In: Proceedings of the 26th International Conference on Neural Information Processing Systems, NIPS’13, pp. 351–359 (2013)

  52. Werro, N., Stormer, H., Meier, A.: Personalized discount—a fuzzy logic approach. In: Proceedings of the 5th IFIP Conference on e-Commerce, E-Business, and E-Government, I3E’05, pp. 375–387 (2005)

  53. Yu, F., Liu, Q., Wu, S., Wang, L., Tan, T.: A dynamic recurrent model for next basket recommendation. In: Proceedings of the 39th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR’16, pp. 729–732 (2016)

  54. Zeiler, M.D.: ADADELTA: an adaptive learning rate method. CoRR arXiv:1212.5701 (2012)

Download references

Author information



Corresponding author

Correspondence to Dietmar Jannach.



Numbers of purchases per user

See the Fig. 6.

Fig. 6

The figure visualizes the purchase frequency distribution in the raw dataset, considering only users who ever made a purchase during the data collection period. The X-axis represents the minimum number of past purchases and on the Y-axis, the percentage of users in the dataset is shown that are above this threshold. For example, about one third of all users made 5 or more purchases

Feature weights for the occasional users

See the Table 14.

Table 14 Results of the statistical feature weight analysis for the occasional user subset

Examined features

See the Tables 15 and 16.

Table 15 The full list of the examined features (see Sect. 2.3) along with their type and a short explanation
Table 16 List of features not considered by DeepPredict

Additional experimental results

See the Tables 17 and 18.

Table 17 Characteristics of the additional Zalando datasets
Table 18 Hit Rate@10 and MRR@10 results for the additional subsets of random and regular Zalando users

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Jannach, D., Ludewig, M. & Lerche, L. Session-based item recommendation in e-commerce: on short-term intents, reminders, trends and discounts. User Model User-Adap Inter 27, 351–392 (2017).

Download citation


  • Recommender systems
  • E-commerce
  • Session-based Recommendation
  • Context-aware Recommendation