Generating Game Levels by Defining Player Experiences
Open accessAbstract
This paper proposes a novel pipeline for generating game levels that elicit predefined emotional experiences from players. Our approach uses evolutionary algorithms alongside data-driven persona agents, predictive emotional models, a PCG parametric level generator, and a newly defined language for the clear and computable definition of player emotional experiences: ExpREx (Experience Regular Expressions). Using these components, we evolve game levels to match the player experience goals specified using the ExpREx language, aiming to create levels that evoke specific emotional experiences for different subsets of players. The efficacy of our method was validated through a user study involving 101 participants, whose continuous annotations of emotional experience were collected and analyzed to assess the congruence between the actual emotional responses elicited and those targeted by our pipeline. We found that 93.73% of the ExpREx goals targeted were also reported by the user study subjects.
AI snapshot
Written from the paper’s full text.
TL;DR
The paper presents ExpREx, a regular-expression-like language for specifying player emotional experiences, and a pipeline that combines persona agents, emotion prediction, parametric level generation, and evolutionary search to create levels matching those specifications. A user study with 101 participants found that 93.73% of targeted ExpREx goals were also reported by players, though the authors frame this as an initial proof-of-concept validation.
Key findings
- In the validation study, 284 of 303 possible ExpREx expression matches were achieved, corresponding to 93.73% accuracy across 101 participants and 3 targeted expressions per player.
- The predictive emotion models reached 72.8% accuracy for Pleasure and 73.1% for Arousal, based on leave-one-out classification of 1-second slices.
- The evolutionary level generator used a population of 100 maps, retained the top 20% as parents, ran for 30 generations, and applied a 10% mutation probability per parameter.
- ExpREx encodes each second as a word containing room/location and Arousal/Pleasure change markers, enabling temporal and spatial emotional patterns to be matched computationally.
Methods
- The test bed was a 2D action-adventure game inspired by classic top-down games; 91 participants, primarily first-year Psychology students, played three hand-designed levels while their actions were recorded, then continuously annotated Pleasure or Arousal after gameplay.
- Persona agents were built by clustering play-traces using Levenshtein distance and evolving parametric agents to mimic each cluster; emotion prediction used random forests trained on gameplay data to predict Pleasure and Arousal changes.
- Levels were evolved with a parametric PCG genome controlling canvas size, path length, rotation and room frequency, corridor and room dimensions, and enemy/coin/health counts; fitness summed matches between five persona agents’ predicted emotional trajectories and the ExpREx expressions.
Limitations
- The authors state the user study is only an initial validation and call for further testing across more levels and different games; they also note the pipeline is not yet robustly evaluated beyond proof of concept.
- The authors state that reliance on ExpREx may limit interpretability and flexibility, may not capture all emotional states, and could benefit from adaptive, context-aware elements and additional experience metrics such as perceived difficulty.
- The authors note that only behavioral play personas were considered, not experience personas, and raise ethical concerns about emotional manipulation, consent, and privacy; evident from the text, the emotion models’ roughly 73% accuracy and the sample of mostly Psychology students may constrain how broadly the 93.73% result generalizes.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 9 cited sources checked · 0 supported · 0 flagged
- Findingown data
In the user study, 284 of the possible 303 expressions were successfully matched, an accuracy of 93.73%.
“In this user study, we were able to successfully match 284 of the possible 303 expressions, which represents an accuracy of 93.73%.”
Quote in the paperSubstantial claim - Findingown data
The predictive model for the Pleasure emotional dimension achieved 72.8% accuracy.
“For the Pleasure dimension, we were able to achieve an accuracy of 72.8%.”
Quote in the paperSubstantial claim - Findingown data
In the user study, 93.73% of the ExpREx goals that the pipeline targeted were also reported by participants.
“We found that 93.73% of the ExpREx goals targeted were also reported by the user study subjects.”
- Also stated in: § 1 Introduction · plain
“The generated level matched the desired emotional experience on 93.73% of the tested ExpREx expressions.”
Quote in the paperSubstantial claim - Also stated in: § 1 Introduction · plain
- Findingown data
The predictive model for the Arousal emotional dimension achieved 73.1% accuracy.
“For the Arousal dimension, we were able to obtain a slightly better 73.1% of accuracy.”
Quote in the paperSubstantial claim - Study design and conductown data
The authors conducted a user study with 101 participants as an initial validation of the approach.
“To validate the effectiveness of our approach, we conducted a user study involving 101 participants.”
- Also stated in: § 8 Conclusion · plain
“An initial validation of the approach was conducted through an empirical study involving 101 participants.”
Quote in the paperSubstantial claim - Also stated in: § 8 Conclusion · plain
- Study design and conductown data
The efficacy of the proposed method was validated through a user study with 101 participants, whose continuous emotional annotations were analyzed for congruence with targeted responses.
“The efficacy of our method was validated through a user study involving 101 participants, whose continuous annotations of emotional experience were collected and analyzed to assess the congruence between the actual emotional responses elicited and those targeted by our pipeline.”
- Also stated in: § 1 Introduction · plain
“the efficacy of our approach is validated through an empirical study involving 101 participants.”
Quote in the paperSubstantial claim - Also stated in: § 1 Introduction · plain
- Premisecitation
Prior work indicates that the Dominance dimension of PAD is difficult to predict with high accuracy.
“We have decided not to use the full PAD model as the Dominance emotional dimension of the model has been shown to be very hard to accurately predict (Fernandes, Lopes, and Prada 2023a).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Fernandes, Lopes, and Prada 2023a
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Study design and conductown data
A generated level was evaluated in the user study by examining player reactions for agreement between intended and actual emotional responses.
“This study tested a generated game level by analyzing player reactions to validate the congruence between intended and actual emotional responses.”
Quote in the paperSubstantial claim - Interpretationown data
The trained predictive model can predict players' emotional responses during new gameplay sessions.
“The trained model enabled us to predict emotional responses of players during new gameplay sessions.”
Quote in the paperSubstantial claim - Interpretationown data
The results showed that the generated level evoked the intended emotional experiences with high accuracy, achieving a match rate of 93.73%.
“The results showed that the generated level was able to evoke the intended emotional experiences with a high degree of accuracy, achieving a match rate of 93.73%.”
Quote in the paperSubstantial claim- Rests on an assumption: The known accuracy of the emotional predictive models (approximately 73%) is a sufficient benchmark for judging the 93.73% match rate as promising.
- Interpretationown data
This finding supports the effectiveness of the proposed methodology in creating game levels that align closely with predefined emotional trajectories.
“This finding supports the effectiveness of the proposed methodology in creating game levels that align closely with the predefined emotional trajectories”
Quote in the paperSubstantial claim- Rests on an assumption: The known accuracy of the emotional predictive models (approximately 73%) is a sufficient benchmark for judging the 93.73% match rate as promising.
- Premisecitation
The emotional predictive models have a known accuracy of approximately 73%.
“This is a promising result for the approach, especially given the known accuracy of both the emotional predictive models to be ≈ 73% (Sec.4.2).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sec.4.2
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationreasoning
Encoding both location and emotional trajectory allows the authors to produce sentences that can be matched against ExpREx expressions.
“By thus encoding the information regarding the location and the emotional trajectory, we are able to create sentences that can then be matched to ExpREx”
Quote in the paperSubstantial claim
- May combine two claims. 11
- Classified as probably not a substantial claim. 19
May combine two claims.
Findingown dataThe evolved levels differ in topology and in the density and placement of in-game objects and enemies.
“These levels differ not only on topology but also on the density and location of the different in game objects and enemies.”
Quote in the paperNot a substantial claimMay combine two claims.
FindingreasoningThe levels generated with unique ExpREx expressions per persona do not guide players along as strict a path as the levels in Fig. 6, allowing greater behavioral choice and different experiences for each persona.
“These generated levels no longer guide the players to follow such a strict path as the levels found in Fig. 6, allowing for a higher degree of behavioural choice and thus allowing each play persona to have a different experience while playing the level.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingnone givenThree different levels were evolved to match sets of ExpREx expressions for five different personas.
“We begin this section by showcasing three different levels that were evolved to match sets of ExpREx expressions for five different personas.”
Quote in the paperNot a substantial claimMay combine two claims.
FindingreasoningThe levels evolved to elicit the same ExpREx patterns across different personas guide players to take similar paths, despite allowing some choice.
“It is not surprising then that these levels, although allowing for some degree of choice, guide the players to take similar paths.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductown dataParticipants were randomly allocated to annotate one of the two emotional dimensions.
“Each participant was randomly assigned one of the dimensions to annotate and provided with a clear explanation of the assigned dimension.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningDividing a generated map into locations gave a per-second location at a coarser granularity than in-game coordinates, which the authors considered more meaningful for the emotional experiences they aimed to encode.
“Having this division, we now had a general location for each second of the game which was of a much lower granularity than the in game coordinates and thus more meaningful for the experiences we wished to encode.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningLevels with higher fitness scores are more likely to survive and propagate in the evolutionary process, resulting in game environments that are emotionally engaging for a diverse range of players and scenarios.
“Levels achieving higher fitness scores are more likely to survive and propagate through the evolutionary process, leading to game environments that are emotionally engaging for a diverse range of players and scenarios.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductown dataModel accuracy was computed as the average classification accuracy of 1-second slices using leave-one-out, with each player's slices classified by a model trained on all other players.
“The accuracy was based on the correct classification of 1 second slices using the “leave one out” method, where we measured the average classification accuracy of the slices from one of the players with a model trained with the data from all the other players.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Interpretationnone givenThe ExpREx language is a novel contribution that uses regular expressions to define and analyze emotional experiences in a clear and computable way.
“The ExpREx language is a novel contribution that leverages regular expressions to define and analyze emotional experiences in a clear and computable manner.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Interpretationnone givenThe genome parameters directly affect the structure and complexity of the generated game maps.
“The genome for the evolutionary algorithm in our level generation process consists of several parameters which directly influence the structure and complexity of the generated game maps.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
InterpretationreasoningExpREx can convert an emotional experience into a string that carries temporal, spatial, and emotional information.
“By sequencing these words, we are able to transform an emotional experience into a string which contains temporal, spacial and emotional information.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
InterpretationreasoningExpREx can describe complex sequences of gameplay experiences by using regular-expression constructs such as quantifiers, character classes, and logical operators.
“By utilizing regex constructs like quantifiers, character classes, and logical operators, ExpREx can effectively describe complex sequences of gameplay experiences.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningExpREx enables precise encoding of complex emotional trajectories over time and space in game levels, making it a robust tool for experience-driven game design.
“This language allows for the precise encoding of complex emotional trajectories over time and space within game levels, providing a robust tool for experience-driven game design.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
PremisecitationPAD represents emotion using three continuous axes: Pleasure, Arousal, and Dominance.
“The PAD model describes emotions along three continuous dimensions: Pleasure (P), Arousal (A), and Dominance (D).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Russell and Mehrabian 1977 source
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationCore Affect maps basic emotions to points on two axes: pleasure-displeasure and activation-deactivation.
“The Core Affect model, on the other hand, represents basic emotional states as points in a two-dimensional space defined by the dimensions of pleasure-displeasure and activation-deactivation.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Russell 2003 source
May combine two claims.
Interpretationown dataOverall, the methodology provides a structured approach to experience-driven game design, enabling the creation of game levels tailored to evoke specific emotional experiences in players.
“Overall, the methodology outlined in this paper provides a structured approach to experience-driven game design, enabling the creation of game levels that are tailored to evoke specific emotional experiences in players.”
Quote in the paperSubstantial claim- Rests on an assumption: The known accuracy of the emotional predictive models (approximately 73%) is a sufficient benchmark for judging the 93.73% match rate as promising.
Classified as probably not a substantial claim.
InterpretationreasoningWhen an emotional experience is represented as a string, regular expressions can be used to encode intended emotional journeys during gameplay.
“Once this information is in string format, we can then leverage the flexibility and precision of regular expressions to encode desired emotional journeys during gameplay.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Premisenone givenPleasure and Arousal are present in both the PAD and Core Affect models and can jointly represent many basic emotional states.
“These dimensions are present in both the PAD model and the core affect model, being able to jointly represent a plethora of basic emotional states.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductcitationEmotional annotations were collected continuously after gameplay using a method similar to prior work by Lopes et al. (2017) and Fernandes et al. (2023a).
“The annotation method was conducted post-gameplay and was continuous, akin to the method described in Lopes et al. (2017) and Fernandes et al. (2023a).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Lopes et al. (2017) source
- cites Fernandes et al. (2023a)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
May combine two claims.
PremisecitationPAD offers a broad framework that captures complex human emotions with just three values, which the authors argue makes it well suited for predicting player experience in games.
“This model provides a comprehensive framework for capturing the complexity of human emotions using only three dimensional values, making it highly suitable for predicting player experiences in games.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Russell and Mehrabian 1977 source
Probably not a substantial claim: may not pass S3 (load-bearing).
InterpretationreasoningUsing ExpREx to define and predict emotional trajectories introduces limitations in the interpretability and flexibility of those definitions.
“The dependence on the ExpREx language to define and predict emotional trajectories introduces limitations related to the interpretability and flexibility of these definitions.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationThe persona agents used in this paper were created using the pipeline described by Fernandes et al. (2023b).
“The persona agents used for this paper were created using this pipeline, of which a brief description can be found in Sec.4.1.”
- Also stated in: § 4.1 The Persona Agent Pipeline · plain
“The persona agents used for this paper followed the same principles and algorithms as the ones described in the work of Fernandes et al. (2023b)”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Fernandes et al. (2023b)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Also stated in: § 4.1 The Persona Agent Pipeline · plain
Probably not a substantial claim: may not pass S3 (load-bearing).
InterpretationreasoningExpREx may not cover all emotional states that occur during gameplay, especially those that do not fit predefined categories.
“While ExpREx provides a computable and structured way to encode emotional responses, it might not encompass all possible emotional states that arise during gameplay, particularly those that do not fit neatly into predefined categories.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationCore Affect concerns the basic structure of emotional experience, which is later interpreted into more complex states.
“This model focuses on the underlying structure of emotional experiences which are then interpreted to form more complex emotional states.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Russell 2003 source
Classified as probably not a substantial claim.
Interpretationown dataExpREx expressions can encode expected emotional trajectories concisely in a computable and flexible form.
“Each of these examples demonstrates how ExpREx can be used to succinctly encode the expected emotional trajectories in a way that is both computable and flexible.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
InterpretationreasoningThe procedural generation method can dynamically create game maps that are varied, interesting, and tailored to enhance player experience through structured yet unpredictable design.
“This procedural method allows for the dynamic creation of game maps that are not only varied and interesting but also tailored to enhance player experience through structured yet unpredictable level design.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim. May combine two claims.
Interpretationnone givenThe results are promising but insufficient beyond a proof of concept; robust evaluation requires further user studies in more levels and different games.
“The results here presented show promise, but they are insufficient beyond a proof of concept and further user studies in several different levels and even several different games are required for a robust evaluation of the pipeline.”
Quote in the paperSubstantial claim- Rests on an assumption: The known accuracy of the emotional predictive models (approximately 73%) is a sufficient benchmark for judging the 93.73% match rate as promising.
May combine two claims.
InterpretationreasoningUsing persona agents, the pipeline can generate play traces and then forecast the expected emotional response for a given level.
“Using the previously described Persona Agents, we could thus generate traces and then predict the expected emotional response of players given a level.”
Quote in the paperSubstantial claim
The paper relies on this without support.
Conventional playtesting and feedback from players tend to be subjective and hard to measure.
“Traditional methods of playtesting and player feedback are often subjective and difficult to quantify.”
The paper relies on this without support.
A key difficulty for automatic level design driven by experience is specifying emotional experiences in a computable and verifiable manner.
“One of the primary challenges in experience based automatic level design is the clear definition of emotional experiences in a way that is both computable and verifiable.”
The paper relies on this without support.
The chosen 2D action-adventure game was simple but allowed more varied player behavior and experience than the simpler games common in PCG studies, so it was deemed a suitable test bed.
“This game was selected due to its simplicity, while at the same time enabling a wider range of player behaviors and experiences than the simpler games often used in PCG research.”
The paper relies on this without support.
The known accuracy of the emotional predictive models (approximately 73%) is a sufficient benchmark for judging the 93.73% match rate as promising.
“This is a promising result for the approach, especially given the known accuracy of both the emotional predictive models to be ≈ 73% (Sec.4.2).”
Cites
Breiman, L. (2001) · Machine Learning · doi:10.1023/a:1010933404324 · Cited by 133,538 · see citing works
Core affect and the psychological construction of emotion.
Russell, J. A. (2003) · Psychological Review · doi:10.1037/0033-295x.110.1.145 · Cited by 6,113 · see citing works
Evidence for a three-factor theory of emotions
Russell, J. A., & Mehrabian, A. (1977) · Journal of Research in Personality · doi:10.1016/0092-6566(77)90037-x · Cited by 1,940 · see citing works
The Inmates are Running the Asylum
Cooper, A. (1999) · Berichte des German Chapter of the ACM · doi:10.1007/978-3-322-99786-9_1 · Cited by 1,855 · see citing works
Search-Based Procedural Content Generation: A Taxonomy and Survey
Togelius, J., Yannakakis, G. N., Stanley, K. O., & Browne, C. (2011) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2011.2148116 · Cited by 706 · see citing works
Procedural content generation for games: A survey
Hendrikx, M., Meijer, S., Van Der Velden, J., & Iosup, A. (2013) · ACM Transactions on Multimedia Computing, Communications, and Applications · doi:10.1145/2422956.2422957 · Cited by 558 · see citing works
Experience-Driven Procedural Content Generation
Yannakakis, G. N., & Togelius, J. (2011) · IEEE Transactions on Affective Computing · doi:10.1109/t-affc.2011.6 · Cited by 536 · see citing works
Artificial Intelligence and Games
Yannakakis, G. N., & Togelius, J. (2018) · doi:10.1007/978-3-319-63519-4 · Cited by 511 · see citing works
Player modeling using self-organization in Tomb Raider: Underworld
Drachen, A., Canossa, A., & Yannakakis, G. N. (2009) · IEEE Symposium on Computational Intelligence and Games (CIG) · doi:10.1109/cig.2009.5286500 · Cited by 288 · see citing works
Ecoffet, A., Huizinga, J., Lehman, J., Stanley, K. O., & Clune, J. (2021) · Nature · doi:10.1038/s41586-020-03157-9 · Cited by 235 · see citing works
RankTrace: Relative and unbounded affect annotation
Lopes, P., Yannakakis, G. N., & Liapis, A. (2017) · 2017 Seventh International Conference on Affective Computing and Intelligent Interaction (ACII) · doi:10.1109/acii.2017.8273594 · Cited by 74 · see citing works
Procedural Personas as Critics for Dungeon Generation
Liapis, A., Holmgård, C., Yannakakis, G. N., & Togelius, J. (2015) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-16549-3_27 · Cited by 50 · see citing works
García‐Sánchez, P., Tonda, A. P., Mora, A. M., et al. (2018) · Knowledge-Based Systems · doi:10.1016/j.knosys.2018.04.030 · Cited by 42 · see citing works
Yannakakis, G. N., & Karpouzis, K. C. (2016) · Socio-affective computing · doi:10.1007/978-3-319-41316-7 · Cited by 41 · see citing works
Generative Personas That Behave and Experience Like Humans
Barthet, M., Khalifa, A., Liapis, A., & Yannakakis, G. N. (2022) · International Conference on the Foundations of Digital Games (FDG) · doi:10.1145/3555858.3555879 · Cited by 30 · see citing works
Evolving models of player decision making: Personas versus clones
Holmgård, C., Liapis, A., Togelius, J., & Yannakakis, G. N. (2015) · Entertainment Computing · doi:10.1016/j.entcom.2015.09.002 · Cited by 20 · see citing works
Towards player-driven procedural content generation
Shaker, N., Yannakakis, G. N., & Togelius, J. (2012) · ACM International Conference on Computing Frontiers (CF) · doi:10.1145/2212908.2212942 · Cited by 15 · see citing works
Agents for Automated User Experience Testing
Fernandes, P. M., Lopes, M. C., & Prada, R. (2021) · IEEE International Conference on Software Testing, Verification and Validation Workshops (ICSTW) · doi:10.1109/icstw52544.2021.00049 · Cited by 11 · see citing works
Self-citationAdapting Procedural Content Generation to Player Personas Through Evolution
Fernandes, P. M., Jorgensen, J., & Poldervaart, N. N. T. G. (2021) · IEEE Symposium Series on Computational Intelligence (SSCI) · doi:10.1109/ssci50451.2021.9659880 · Cited by 6 · see citing works
Self-citationEmoSTL: Formal Spatial-Temporal Verification of Emotion Specifications in Computer Games
Ansari, S. G., Prasetya, I. S. W. B., Dastani, M. M., et al. (2024) · IEEE Conference on Software Testing, Verification and Validation (ICST) · doi:10.1109/icst60714.2024.00011 · Cited by 2 · see citing works
Data Driven Agents for User Experience Testing
Fernandes, P. M., Lopes, M. C., & Prada, R. (2024) · Communications in computer and information science · doi:10.1007/978-3-031-51452-4_1 · Cited by 1 · see citing works
Self-citationPersona Agents from Playtraces and Emotion
Fernandes, P. M., Lopes, M. C., & Prada, R. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment · doi:10.1609/aiide.v19i1.27517 · Cited by 1 · see citing works
Self-citationThe Arousal video Game AnnotatIoN (AGAIN) Dataset
Melhart, D., Liapis, A., & Yannakakis, G. N. (2021) · arXiv · doi:10.48550/arxiv.2104.02643
Flagged first, then by citations
Research works printed but not linked: 1
- [24]24. Yannakakis, G. N.; and Togelius, J. 2018. Artificial Intelli- gence and Games. In Springer.
Cited by
Sorted by citations
Player Modeling and Large Language Models: An Interaction-Based Approach
Souza, C. H. R., Berretta, L. D. O., & Carvalho, S. T. (2025) · Simpósio Brasileiro de Jogos e Entretenimento Digital (SBGames) · doi:10.5753/sbgames.2025.10174 · Cited by 1 · see citing works
“Player Modeling involves creating computational models that describe players’ states, predict their future behaviors, and identify their preferences [Tagliaro 2022, Fernandes et al. 2025, Mortazavi et al. 2024].”
InfluentialA dynamic features method for retrieval and emotional polarity of digital media images
Xin, X., & Arshad, M. (2025) · PeerJ Computer Science · doi:10.7717/peerj-cs.3265
High Dimensional Procedural Content Generation
Xu, K., & Verbrugge, C. (2026) · International Conference on the Foundations of Digital Games (FDG) · doi:10.1145/3815598.3815606
- Authors
- Fernandes, P. M., Lopes, M., & Prada, R.
- Affiliations
- Instituto de Engenharia de Sistemas e Computadores Investigação e Desenvolvimento (PT)
- Venue
- Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment
- Published
- 2024-11-15
- Volume
- 20
- Issue
- 1
- Pages
- 179-188
- ISSN
- 2326-909X (ISSN-L)
- 2334-0924
- OpenAlex
- W4404445400
- Cited by
- 3
- References
- 20
- Citations by year
- 2026: 1 · 2025: 2
- FWCI
- 0.87
- Type
- conference paper
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (publishedVersion)
- Funders
- Fundação para a Ciência e a Tecnologia (UIDB/50021/2020)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Social Sciences › Sociology and Political Science › Digital Games and Media
- Physical Sciences › Computer Science › Artificial Intelligence › Evolutionary Algorithms and Applications
- Keywords
- game level generation, procedural content generation, emotion induction, evolutionary algorithms, user study
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-01.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-01.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
