Pith. sign in

REVIEW 4 major objections 6 minor 53 references

Modeling Musical Genre Trajectories through Pathlet Learning

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that recurring short genre sequences, called pathlets, learned from user listening trajectories, predict which genres appear and disappear in a user's taste, beating standard baselines while remaining interpretable.

desk verdict A worthwhile domain adaptation of pathlet learning with a valuable data release, but the reported quantitative win lacks a non-pathlet control and is not yet proven. read the letter →

arxiv 2505.03480 v1 pith:IQQ6EV6D submitted 2025-05-06 cs.IR cs.LGcs.MM

Lilian Marey , Charlotte Laclau , Bruno Sguerra , Tiphaine Viard , Manuel Moussallam This is my paper · ORCID
classification cs.IRcs.LGcs.MM
keywords pathletlearningdictionarymusicalgenretrajectoryembeddingsappearanceanddisappearanceinterpretablemusicrecommendationuserpreferencesevolution
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to show that dictionary learning can be moved from road networks to music streaming, turning each user's genre listening history into short recurring patterns called pathlets. Pathlets are sub-sequences of co-listened genres ranked by how important they are to the user, and the paper claims that these patterns carry enough information to anticipate whether a genre will appear, persist, or disappear in the next month of listening. If true, this gives an interpretable alternative to black-box taste models: each embedding coordinate corresponds to a readable listening pattern rather than a latent dimension. The paper also releases a 17-month, 2000-user genre-labeled listening dataset and reports that its method lowers prediction error on genre allocation and sharply improves detection of newly appearing genres.

What carries the argument

The load-bearing object is the co-listening history vector $\eta^k_{u,g}$ (Eq. 3), which counts, per time window, how often each other genre is listened to immediately before or after genre $g$ in user $u$'s sequence. From these counts the method samples a rank trajectory—each step is the rank of a co-listened genre, with the target genre at rank 0—then learns a dictionary of the most frequent sub-paths (the pathlets) by minimizing a sparse reconstruction loss $ \frac{1}{2}\|P - D_0\alpha\|_F^2 + \lambda\|\alpha\|_1$. The trajectory embedding is produced not by $\alpha$ directly but by greedy longest-pathlet segmentation, which avoids redundant activations. These embeddings feed random-forest classifiers that flag likely appearances and disappearances; the flags then adjust the strong 'Previous window' baseline into the Plug-Previous predictor.

What would settle it

Rebuild the trajectory dictionaries using only genre pairs that sit two or more positions apart in the listening history instead of adjacent pairs; if the Plug-Previous gain over the Previous baseline on Deezer disappears or reverses, the first-order adjacency assumption in Eq. 3 is what carries the result. A second check: shuffle the order of tracks within each time window before computing $\eta^k_{u,g}$; order-independent predictions would indicate the pathlets add nothing beyond genre-allocation frequencies.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that pathlet-based trajectory embeddings preserve signal about the dynamics of genre emergence and abandonment. The authors build, for each user and each genre, a rank-encoded trajectory of which genres accompany that genre over time, learn a sparse dictionary of recurring sub-paths, encode trajectories by matching longest pathlets, and feed those embeddings to simple classifiers that decide whether a genre will appear or disappear in the next window. On the Deezer dataset this 'Plug-Previous' scheme lowers the average total variation of predicted genre allocations from 0.366 to 0.348 and raises the New classes AUC from 0.660 to 0.827, while on Last-fm it raises New classes from 0.681 to 0.999. The paper further claims a qualitative payoff: inertial pathlets (where the target genre accompanies itself) correlate with genre retention, and the joint presence of rock and metal in a hard-rock trajectory is associated with hard rock persisting, illustrating genre competition that an analyst can read directly from the learned patterns.

Load-bearing premise

The construction assumes that which genres accompany a genre immediately before or after it in a listening sequence (plus a fallback to overall genre allocation when no co-listened genre exists) is enough to capture why that genre appears or disappears; if emergence and decline actually depend on longer-range listening context, the pathlet signal collapses.

Editorial extensions

If this is right

  • Genre appearance and disappearance can be treated as predictable, labeled events rather than unexplained drift, because explicit embeddings trained on adjacent co-listening beat strong baselines at both subtasks.
  • Interpretable taste models become feasible for streaming platforms: an embedding coordinate is a readable pattern, so recommendations can be explained as 'this user typically moves toward genre X via co-listening with Y.'
  • The released 17-month, 2000-user genre-labeled dataset gives other researchers a common benchmark for trajectory and taste-evolution models.
  • The ranked-trajectory trick (replacing genre identities by user-level ranks) suggests pathlet learning can transfer between users even when the underlying genres differ, which matters for cold-start and cross-domain taste modeling.
  • Qualitative genre-competition findings, such as rock plus metal jointly reinforcing hard rock, are direct corollaries of the learned pathlet correlations and can be checked against independent listening data.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the paper only tests adjacent co-listening, a natural extension is to let the pathlet dictionary length itself reveal which context scale matters—if longer pathlets systematically activate more for genre emergence, the first-order assumption is a restriction rather than a discovered fact.
  • The trajectory-embedding principle may transfer beyond music: any domain with dense sequential consumption (podcasts, short video, news hits) could use the same rank-encoding and longest-match segmentation, provided interaction density is high enough.
  • The finding that both popular and niche genres arise from diversified trajectories suggests a testable recommendation policy: deliberately diversifying a user's current context should increase the appearance rate of new genres, which could be validated with a small randomized listening experiment.
  • The reliance on a fallback to genre-allocation sampling when no co-listened genre exists means the method's success could partly be carried by that fallback rather than by the pathlets; separating these contributions would clarify where the predictive signal actually lives.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes a pathlet-learning framework to model users' musical genre trajectories over time. Listen histories are converted into genre allocation vectors per time window, and for each user-genre pair a trajectory of co-listened genre ranks is sampled from first-order adjacency counts (with a fallback to genre allocation). A dictionary of recurrent sub-trajectories (pathlets) is learned via an ℓ1-regularized reconstruction, and trajectory embeddings are built by greedily decomposing each trajectory into the longest matching pathlets. These embeddings are then used as features in random-forest classifiers that predict genre appearance/disappearance, and the output is plugged into the Previous (persistence) baseline to form Plug-Previous. The method is evaluated on a new Deezer dataset and on Last.fm using ATV, Plus-minus, and New classes metrics, with qualitative analyses of pathlets and genre interaction graphs. The Deezer dataset and code are released.

Significance. If the empirical claims hold, this is a useful interpretable alternative to black-box models for studying preference dynamics, and the released Deezer dataset (2,000 users, 17 months, genre-labeled histories) is a valuable community resource. The paper's strengths include the concrete dataset release, the reproducible code, the explicit dictionary-learning formulation with interpretable atoms, and the qualitative pathlet analyses that connect predictive signals to plausible genre-interaction mechanisms. The central quantitative claim, however, is not yet established: the ATV improvement over the Previous baseline is small, appears only on Deezer, and is not shown to be attributable to the pathlet embeddings themselves.

major comments (4)
  1. [Section 6.1, Table 1] The central quantitative claim is that Plug-Previous improves ATV over Previous on Deezer (0.348 vs 0.366), but the table reports point estimates without error bars, significance tests, or repeated seeded runs, despite the random trajectory sampling in Section 4.2 and the stochasticity of the random-forest classifiers. Moreover, the Last.fm result shows Plug-Previous underperforming Previous (0.506 vs 0.491), which contradicts the general claim. Please provide variance estimates, confidence intervals, or significance tests across multiple random seeds, and discuss the Last.fm discrepancy.
  2. [Section 5.2, Section 6.1] The reported Deezer ATV gain is not demonstrated to come from the pathlet embeddings. Plug-Previous starts from the Previous prediction and only adjusts entries in A+ and A− using random-forest classifiers; a control using non-pathlet features (e.g., historical mean allocation, candidate-set membership, or raw averaged co-listening counts) is missing. Without this ablation, the improvement could be due to the classification machinery rather than to the pathlet representation. Please add such non-pathlet feature controls.
  3. [Section 5.2] The Preference Transition Model (PTM) from [33] is the original model for the evaluation task, yet its results are omitted with only the statement that it 'performed unexpectedly poorly' and that lack of code hindered reproduction. This is a load-bearing comparison; please report the reproduced PTM scores (or a detailed description of the failure), so readers can verify the baseline setup and the relevance of the comparison.
  4. [Section 4.2, Eq. (3)] The trajectory construction relies exclusively on first-order adjacency in the co-listening vector η^k_{u,g}, with a fallback to genre-allocation sampling when no co-listen occurs. If genre emergence or decline depends on longer-range or non-adjacent listening context, the pathlet dictionaries will miss that signal. The paper should demonstrate the sensitivity of the downstream predictions to this first-order assumption, e.g., by comparing with trajectories built on higher-order co-occurrence windows or by quantifying how often the fallback path is active.
minor comments (6)
  1. [Section 5.4] The Last.fm dataset description states '5 three-months time windows from August 2007 to January 2019,' which is temporally inconsistent; likely the end year is a typo (e.g., 2008 or 2009). Please correct it.
  2. [Table 1 footnote] The asterisk note saying 'results are computed with 1 time window shift for metric validity' is vague; please clarify exactly what was shifted and why this adjustment is needed for the Plus-minus and New classes metrics.
  3. [Section 5.3, Figure 4] The sensitivity analysis for λ is performed on the same datasets used for final evaluation, so λ=0.0025 is effectively selected on the test data; please acknowledge this and, if possible, use an inner validation split or report results under neighboring λ values.
  4. [Section 4.4] The trajectory embedding algorithm is described as 'iteratively identifies the longest matching pathlet,' but the matching rule (e.g., tie-breaking, whether overlapping pathlets are allowed) and the time complexity are not specified; please give a precise algorithmic description.
  5. [Section 4.2] When no co-listened genre exists, the text says the sample is drawn 'from genre allocation values from X^k_{u,g},' but the exact sampling distribution (normalized proportions, or uniform over heard genres?) is not defined; please state it.
  6. [Typos and grammar] Several typos and stylistic issues should be fixed: 'prenseted' (Section 5.4), 'the results of our Previous-plug prove' (Section 6.1), 'difficult reproduction' (Section 5.2), and the caption 'Sensitive Analysis' in Figure 4.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the target Y is not used in dictionary construction, and the prediction pipeline is temporally separated from the final window.

full rationale

The derivation chain is self-contained and does not reduce to its own inputs. The final-window target Y is explicitly excluded from trajectory construction: trajectories are built from the first K−1 time windows (Section 4.2), and the candidate sets A+ and A− are defined using only Xprev and earlier windows, not the final allocation. The dictionaries and pathlet embeddings are learned from these pre-target trajectories, and the Random Forest classifiers are trained on the first K−1 windows and evaluated on the final window (Section 5.4). Plug-Previous starts from the Previous baseline and modifies only candidate pairs, but the modification is driven by classifier outputs that are not derived from Y; assigning a historical average or zero is a decision rule, not a fitted parameter renamed as a prediction. Hyperparameters such as lambda and dictionary size are selected on reconstruction quality of unsupervised dictionaries, not on ATV or classification performance against Y, so there is no fitted-input-called-prediction circularity. The paper's self-citations (e.g., [1], [5], [14], [19], [42]) appear only in related-work or methodological background and are not load-bearing for the central claim. No uniqueness theorem or ansatz is imported from the authors' prior work to force the choice of pathlets. The absence of a non-pathlet control for the Plug-Previous gain is a legitimate experimental limitation, but it is an attribution gap, not a circular derivation: the reported improvement does not equal an input by construction. Accordingly, no circular step meets the evidence bar of quoting a specific equation or reduction that equates the prediction with the input.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The method rests on domain-specific assumptions about co-listening adjacency and rank encoding, plus tuning of λ and several size parameters. No new physical or conceptual entities are introduced.

free parameters (5)
  • λ (sparsity regularization) = 0.0025
    Selected via sensitivity analysis in Figure 4 on the same Deezer and Last.fm datasets; controls the balance between reconstruction and sparsity in Eq. 4.
  • Number of sampled trajectories per candidate = 1,000 per candidate; 5,000 selected
    Chosen from reconstruction-error curve in Figure 3; affects the pathlet dictionary and embeddings.
  • Number of candidate pathlets = 10,000
    Chosen for memory and interpretability constraints (Section 5.3), not derived from the data or theory.
  • Maximum pathlet size = < 10
    Set to keep pathlets manageable; no sensitivity analysis reported.
  • Random forest hyperparameters = unspecified
    Defaults are not stated, making the classification results hard to reproduce exactly.
assumptions (5)
  • domain assumption Co-listening counts computed from genres immediately before or after g within a listening session capture the interactions that drive genre appearance and disappearance.
    The trajectory definitions in Eq. 3 and Section 4.2 assume first-order adjacency is the relevant signal.
  • domain assumption When no co-listened genre is available, sampling from the genre allocation X is a valid substitute.
    Section 4.2 uses this fallback to complete trajectories.
  • domain assumption Ranking genres by user listening frequency and encoding trajectories as ranks preserves the structure needed for pathlet learning.
    Section 4.2: 'This method helps to reduce sparsity, and suggests that musical genres can assume interchangeable roles'.
  • standard math The dictionary learning objective (Eq. 4) with ℓ1 regularization and the subsequent greedy segmentation produce embeddings that faithfully represent trajectory content.
    Uses convex optimization and greedy matching, which are standard tools.
  • standard math The graph induced by the trajectory set P is a valid representation of genre transitions.
    Section 4.3 defines the graph as the smallest directed graph in which all paths are valid.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Modeling Musical Genre Trajectories through Pathlet Learning." pith.science (2026). https://pith.science/paper/IQQ6EV6D

@misc{pith2026250503480,
  author       = {Pith},
  title        = {Pith review of: Modeling Musical Genre Trajectories through Pathlet Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IQQ6EV6D}},
  note         = {Machine review of arXiv:2505.03480}
}
read the original abstract

The increasing availability of user data on music streaming platforms opens up new possibilities for analyzing music consumption. However, understanding the evolution of user preferences remains a complex challenge, particularly as their musical tastes change over time. This paper uses the dictionary learning paradigm to model user trajectories across different musical genres. We define a new framework that captures recurring patterns in genre trajectories, called pathlets, enabling the creation of comprehensible trajectory embeddings. We show that pathlet learning reveals relevant listening patterns that can be analyzed both qualitatively and quantitatively. This work improves our understanding of users' interactions with music and opens up avenues of research into user behavior and fostering diversity in recommender systems. A dataset of 2000 user histories tagged by genre over 17 months, supplied by Deezer (a leading music streaming company), is also released with the code.

Figures

Figures reproduced from arXiv: 2505.03480 by the authors.

Figure 1
Figure 1. Our methodology for users preferences modeling. Music genre trajectories are extracted from user listening [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Users’ genre listening variation types between 2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. determined that a value of λ = 0.0025 balances high reconstruction scores with embedding expressiveness in our PL model. We optimize the function presented in Equation 4 using the Adam optimizer with a learning rate of 0.01 and apply an early stopping criterion based on loss stagnation over 5 epochs. For the appearance and dis￾appearance classification tasks, we utilize random forest classifiers, training them on th… view at source ↗
Figures from the paper (3 more)
Figure 3
Figure 3. Figure 3: Reconstruction error of X over the number of sampled trajectories. 5.4 Datasets We apply our method on Deezer dataset, which contains 24M interactions between 2, 000 users and 350 music gen￾res over 17 one-month time windows from January 2022 to June 2023. To expand ou…
Figure 5
Figure 5. Figure 5: Pathlet interpretations : interactions between genres (top) and pathlet diversity (bottom) [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Pathlets correlation with appearance (top) and [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 52 canonical work pages

  1. [33]

    Where to next? a dynamic model of user preferences

    Francesco Sanna Passino, Lucas Maystre, Dmitrii Moor, Ashton Anderson, and Mounia Lalmas. Where to next? a dynamic model of user preferences. In Proceedings of the Web Conference 2021, pages 3210– 3220, 2021

    work page 2021

  2. [1]

    Uplifting interviews in social science with individual data visualization: The case of music listening

    Robin Cura, Am´ elie Beaumont, Jean-Samuel Beuscart, Samuel Coavoux, No´ e Latreille de Fozi` eres, Brenda Le Bigot, Yann Renisio, Manuel Moussallam, and Thomas Louail. Uplifting interviews in social science with individual data visualization: The case of music listening. In CHI Conference on Human Factors in Computing Systems Extended Abstracts , pages 1–9, 2022

    work page 2022

  3. [2]

    Taste over time: The tem- poral dynamics of user preferences

    Joshua L Moore, Shuo Chen, Douglas R Turnbull, and Thorsten Joachims. Taste over time: The tem- poral dynamics of user preferences. In ISMIR, vol- ume 13, page 406, 2013

    work page 2013

  4. [3]

    Tastepaths: Enabling deeper exploration and understanding of personal preferences in recom- mender systems

    Savvas Petridis, Nediyana Daskalova, Sarah Men- nicken, Samuel F Way, Paul Lamere, and Jennifer Thom. Tastepaths: Enabling deeper exploration and understanding of personal preferences in recom- mender systems. In Proceedings of the 27th Inter- national Conference on Intelligent User Interfaces , pages 120–133, 2022

    work page 2022

  5. [4]

    Computing taste: Algorithms and the 9 makers of music recommendation

    Nick Seaver. Computing taste: Algorithms and the 9 makers of music recommendation . University of Chicago Press, 2022

    work page 2022

  6. [5]

    Follow the guides: disentangling human and algorithmic cu- ration in online music consumption

    Quentin Villermet, J´ er´ emie Poiroux, Manuel Mous- sallam, Thomas Louail, and Camille Roth. Follow the guides: disentangling human and algorithmic cu- ration in online music consumption. In Proceedings of the 15th ACM Conference on Recommender Systems, pages 380–389, 2021

    work page 2021

  7. [6]

    Distinction a social critique of the judgement of taste

    Pierre Bourdieu. Distinction a social critique of the judgement of taste. In Inequality, pages 287–318. Routledge, 2018

    work page 2018

  8. [7]

    Music in everyday life

    Tia DeNora. Music in everyday life . Cambridge uni- versity press, 2000

    work page 2000

Show all 53 references
  1. [8]

    Why music matters

    David Hesmondhalgh. Why music matters. John Wi- ley & Sons, 2013

    2013

  2. [9]

    Music and discov- ery

    Tom McCourt and Nabeel Zuberi. Music and discov- ery. Popular Communication, 14(3):123–126, 2016

    2016

  3. [10]

    Discovering music at sofar sounds: Sur- prise, attachment, and the fan–artist relationship

    Lo¨ ıc Riom. Discovering music at sofar sounds: Sur- prise, attachment, and the fan–artist relationship. Popular Music, Technology, and the Changing Media Ecosystem: From Cassettes to Stream, pages 201–216, 2020

    2020

  4. [11]

    Current challenges and visions in music recommender systems research

    Markus Schedl, Hamed Zamani, Ching-Wei Chen, Yashar Deldjoo, and Mehdi Elahi. Current challenges and visions in music recommender systems research. International Journal of Multimedia Information Re- trieval, 7:95–116, 2018

    2018

  5. [12]

    Neither black nor box: Ways of know- ing algorithms

    Taina Bucher. Neither black nor box: Ways of know- ing algorithms. Innovative methods in media and communication research, pages 81–98, 2016

    2016

  6. [13]

    Traces of globalization in online music consumption patterns and results of recommendation algorithms

    Oleg Lesota, Emilia Parada-Cabaleiro, Stefan Brandl, Elisabeth Lex, Navid Rekabsaz, and Markus Schedl. Traces of globalization in online music consumption patterns and results of recommendation algorithms. In ISMIR, pages 291–297, 2022

    2022

  7. [14]

    Do recommender systems promote local music? a reproducibility study using music streaming data

    Kristina Matrosova, Lilian Marey, Guillaume Salha- Galvan, Thomas Louail, Olivier Bodini, and Manuel Moussallam. Do recommender systems promote local music? a reproducibility study using music streaming data. In Proceedings of the 18th ACM Conference on Recommender Systems, pa...

    2024

  8. [15]

    Putting popularity bias mitigation to the test: A user-centric evaluation in music recommenders

    Robin Ungruh, Karlijn Dinnissen, Anja Volk, Maria Soledad Pera, and Hanna Hauptmann. Putting popularity bias mitigation to the test: A user-centric evaluation in music recommenders. In Proceedings of the 18th ACM Conference on Recommender Systems, pages 169–178, 2024

    2024

  9. [16]

    Diversity of what? on the different conceptualizations of diversity in recommender sys- tems

    Sanne Vrijenhoek, Savvina Daniil, Jorden Sandel, and Laura Hollink. Diversity of what? on the different conceptualizations of diversity in recommender sys- tems. In The 2024 ACM Conference on Fairness, Ac- countability, and Transparency, pages 573–584, 2024

    2024

  10. [17]

    Recommend me? designing fairness metrics with providers

    Jessie J Smith, Aishwarya Satwani, Robin Burke, and Casey Fiesler. Recommend me? designing fairness metrics with providers. In The 2024 ACM Conference on Fairness, Accountability, and Transparency, pages 2389–2399, 2024

    2024

  11. [18]

    User perception of differences in recommender algorithms

    Michael D Ekstrand, F Maxwell Harper, Martijn C Willemsen, and Joseph A Konstan. User perception of differences in recommender algorithms. In Pro- ceedings of the 8th ACM Conference on Recommender systems, pages 161–168, 2014

    2014

  12. [19]

    Explainability in music recommender systems

    Darius Afchar, Alessandro Melchiorre, Markus Schedl, Romain Hennequin, Elena Epure, and Manuel Moussallam. Explainability in music recommender systems. AI Magazine, 43(2):190–208, 2022

    2022

  13. [20]

    Large-scale parallel collaborative fil- tering for the netflix prize

    Yunhong Zhou, Dennis Wilkinson, Robert Schreiber, and Rong Pan. Large-scale parallel collaborative fil- tering for the netflix prize. In Algorithmic Aspects in Information and Management: 4th International Conference, AAIM 2008, Shanghai, China, June 23- 25, 2008. Proceedings ...

    2008

  14. [21]

    Col- laborative filtering for implicit feedback datasets

    Yifan Hu, Yehuda Koren, and Chris Volinsky. Col- laborative filtering for implicit feedback datasets. In 2008 Eighth IEEE international conference on data mining, pages 263–272. Ieee, 2008

    2008

  15. [22]

    Alternating least squares for personalized ranking

    G´ abor Tak´ acs and Domonkos Tikk. Alternating least squares for personalized ranking. In Proceedings of the sixth ACM conference on Recommender systems , pages 83–90, 2012

    2012

  16. [23]

    Using content-based filtering for recommendation

    Robin Van Meteren and Maarten Van Someren. Using content-based filtering for recommendation. In Pro- ceedings of the machine learning in the new informa- tion age: MLnet/ECML2000 workshop , volume 30, pages 47–56. Barcelona, 2000

    2000

  17. [24]

    Hybrid recommender systems: Survey and experiments

    Robin Burke. Hybrid recommender systems: Survey and experiments. User modeling and user-adapted in- teraction, 12:331–370, 2002

    2002

  18. [25]

    Modeling user preferences in recommender systems: A classification framework for explicit and implicit user feedback

    Gawesh Jawaheer, Peter Weller, and Patty Kostkova. Modeling user preferences in recommender systems: A classification framework for explicit and implicit user feedback. ACM Transactions on Interactive In- telligent Systems (TiiS) , 4(2):1–26, 2014

    2014

  19. [26]

    Neural collabora- tive filtering

    Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu, and Tat-Seng Chua. Neural collabora- tive filtering. In Proceedings of the 26th international conference on world wide web , pages 173–182, 2017. 10

    2017

  20. [27]

    Recurrent neu- ral networks

    Larry R Medsker, Lakhmi Jain, et al. Recurrent neu- ral networks. Design and Applications , 5(64-67):2, 2001

    2001

  21. [28]

    Long short-term mem- ory

    Alex Graves and Alex Graves. Long short-term mem- ory. Supervised sequence labelling with recurrent neu- ral networks, pages 37–45, 2012

    2012

  22. [29]

    Transformers meet act-r: Repeat-aware and sequential listening session recommendation

    Viet-Anh Tran, Guillaume Salha-Galvan, Bruno Sguerra, and Romain Hennequin. Transformers meet act-r: Repeat-aware and sequential listening session recommendation. In Proceedings of the 18th ACM Conference on Recommender Systems , pages 486– 496, 2024

    2024

  23. [30]

    Long and short- term recommendations with recurrent neural net- works

    Robin Devooght and Hugues Bersini. Long and short- term recommendations with recurrent neural net- works. In Proceedings of the 25th conference on user modeling, adaptation and personalization , pages 13– 21, 2017

    2017

  24. [31]

    Music classification based on genre using lstm

    S Deepak and BG Prasad. Music classification based on genre using lstm. In 2020 Second International Conference on Inventive Research in Computing Ap- plications (ICIRCA), pages 985–991. IEEE, 2020

    2020

  25. [32]

    Embedding-based next song recommendation for playlists

    Rapha¨ el Romero and Tijl De Bie. Embedding-based next song recommendation for playlists. In European Symposium on Artificial Neural Networks, Computa- tional Intelligence and Machine Learning , 2022

    2022

  26. [34]

    Learning from incomplete ratings using non-negative matrix factorization

    Sheng Zhang, Weihong Wang, James Ford, and Fil- lia Makedon. Learning from incomplete ratings using non-negative matrix factorization. In Proceedings of the 2006 SIAM international conference on data min- ing, pages 549–553. SIAM, 2006

    2006

  27. [35]

    Scalable recommendation with hierarchical poisson factorization

    Prem Gopalan, Jake M Hofman, and David M Blei. Scalable recommendation with hierarchical poisson factorization. In UAI, pages 326–335, 2015

    2015

  28. [36]

    A survey on knowledge graph-based recommender systems

    Qingyu Guo, Fuzhen Zhuang, Chuan Qin, Hengshu Zhu, Xing Xie, Hui Xiong, and Qing He. A survey on knowledge graph-based recommender systems. IEEE Transactions on Knowledge and Data Engineering, 34 (8):3549–3568, 2020

    2020

  29. [37]

    Explainable sequential recommendation using knowledge graphs

    Hao Hou and Chongyang Shi. Explainable sequential recommendation using knowledge graphs. In Proceed- ings of the 5th International Conference on Frontiers of Educational Technologies, pages 53–57, 2019

    2019

  30. [38]

    A generic reinforced explainable framework with knowledge graph for session-based recommendation

    Huizi Wu, Hui Fang, Zhu Sun, Cong Geng, Xinyu Kong, and Yew-Soon Ong. A generic reinforced explainable framework with knowledge graph for session-based recommendation. In 2023 IEEE 39th International Conference on Data Engineering (ICDE), pages 1260–1272. IEEE, 2023

    2023

  31. [39]

    Online dictionary learning for sparse coding

    Julien Mairal, Francis Bach, Jean Ponce, and Guillermo Sapiro. Online dictionary learning for sparse coding. In Proceedings of the 26th annual inter- national conference on machine learning , pages 689– 696, 2009

    2009

  32. [40]

    Efficient convolutional sparse cod- ing

    Brendt Wohlberg. Efficient convolutional sparse cod- ing. In 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages 7173–7177. IEEE, 2014

    2014

  33. [41]

    Shift-invariance sparse coding for audio classification

    Roger Grosse, Rajat Raina, Helen Kwong, and An- drew Y Ng. Shift-invariance sparse coding for audio classification. arXiv preprint arXiv:1206.5241 , 2012

    2012 arXiv

  34. [42]

    Modeling activity-driven music listening with pace

    Lilian Marey, Bruno Sguerra, and Manuel Moussal- lam. Modeling activity-driven music listening with pace. In Proceedings of the 2024 Conference on Hu- man Information Interaction and Retrieval , pages 346–351, 2024

    2024

  35. [43]

    Pathletrl: Trajec- tory pathlet dictionary construction using reinforce- ment learning

    Gian Alix and Manos Papagelis. Pathletrl: Trajec- tory pathlet dictionary construction using reinforce- ment learning. In Proceedings of the 31st ACM Inter- national Conference on Advances in Geographic In- formation Systems, pages 1–12, 2023

    2023

  36. [44]

    Ex- plainable trajectory representation through dictio- nary learning

    Yuanbo Tang, Zhiyuan Peng, and Yang Li. Ex- plainable trajectory representation through dictio- nary learning. In Proceedings of the 31st ACM Inter- national Conference on Advances in Geographic In- formation Systems, pages 1–4, 2023

    2023

  37. [45]

    Pathlet learning for compress- ing and planning trajectories

    Chen Chen, Hao Su, Qixing Huang, Lin Zhang, and Leonidas Guibas. Pathlet learning for compress- ing and planning trajectories. In Proceedings of the 21st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems , pages 392–395, 2013

    2013

  38. [46]

    Honglak Lee, Alexis Battle, Rajat Raina, and An- drew Y. Ng. Efficient sparse coding algorithms. In Proceedings of the 19th International Conference on Neural Information Processing Systems , NIPS’06, page 801–808, Cambridge, MA, USA, 2006. MIT Press

    2006

  39. [47]

    Image compres- sion using learned dictionaries by rls-dla and com- pared with k-svd

    Karl Skretting and Kjersti Engan. Image compres- sion using learned dictionaries by rls-dla and com- pared with k-svd. In 2011 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP), pages 1517–1520. IEEE, 2011

    2011

  40. [48]

    Image denoising via sparse and redundant representations over learned dictionaries

    Michael Elad and Michal Aharon. Image denoising via sparse and redundant representations over learned dictionaries. IEEE Transactions on Image processing, 15(12):3736–3745, 2006. 11

    2006

  41. [49]

    Anomaly detection in mixed telemetry data using a sparse representation and dictionary learning

    Barbara Pilastre, Loic Boussouf, St´ ephane d’Escrivan, and Jean-Yves Tourneret. Anomaly detection in mixed telemetry data using a sparse representation and dictionary learning. Signal Processing, 168:107320, 2020

    2020

  42. [50]

    Dictionary learning and sparse coding for unsupervised cluster- ing

    Pablo Sprechmann and Guillermo Sapiro. Dictionary learning and sparse coding for unsupervised cluster- ing. In 2010 IEEE international conference on acous- tics, speech and signal processing , pages 2042–2045. IEEE, 2010

    2010

  43. [51]

    Anomaly detection in self-organizing industrial systems using pathlets

    Marie Kiermeier, Martin Werner, Claudia Linnhoff- Popien, Horst Sauer, and Jan Wieghardt. Anomaly detection in self-organizing industrial systems using pathlets. In 2017 IEEE International Conference on Industrial Technology (ICIT) , pages 1226–1231. IEEE, 2017

    2017

  44. [52]

    Music recommendation and discovery in the long tail

    `Oscar Celma Herrada et al. Music recommendation and discovery in the long tail . Universitat Pompeu Fabra, 2009

    2009

  45. [53]

    The unfairness of popularity bias in music recommen- dation: A reproducibility study

    Dominik Kowald, Markus Schedl, and Elisabeth Lex. The unfairness of popularity bias in music recommen- dation: A reproducibility study. In Advances in Infor- mation Retrieval: 42nd European Conference on IR Research, ECIR 2020, Lisbon, Portugal, April 14–17, 2020, Proceedings,...

    2020

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.