Research
Publications
Accountability Adaptation Algorithmic Bias Algorithmic Fairness Algorithmic Transparency Artificial Intelligence Collaborative Learning crowdsourcing Diversity Education Ethics Explainability Information Retrieval Information Studies Information Systems mixed reality Personalization Recommender Systems Virtual Environments
Paul D Clough, Jahna Otterbacher Democratizing AI: From Theory to Practice Journal Article Forthcoming The Handbook of Research on Artificial Intelligence, Innovation, and Entrepreneurship. London: Edward Elgar Publishing., Forthcoming. Abstract | BibTeX | Tags: Artificial Intelligence @article{Clough2022, title = {Democratizing AI: From Theory to Practice}, author = {Paul D Clough, Jahna Otterbacher }, year = {2022}, date = {2022-01-31}, journal = {The Handbook of Research on Artificial Intelligence, Innovation, and Entrepreneurship. London: Edward Elgar Publishing.}, abstract = {We are witnessing a movement towards Democratizing AI, with a wide variety of tools, platforms and data sources becoming accessible to more people. This movement is said to be fueling innovation, extending the capabilities of individuals and organizations, by making the creation and application of AI easier. However, beyond the hype, there is a need to understand what this trend means for various stakeholders. Through the lens of socio-political democracy, this chapter examines the democratization of AI. We find that the present state of the “AI Democracy” maps onto only one of three elements of a democracy. Current efforts focus primarily on providing people with the tools and technical infrastructure needed to participate in AI, but not in protecting their freedoms and access to social benefits, which are the other core elements of democracy. We discuss the possibilities for realizing a broader AI democracy, along with the anticipated challenges.}, keywords = {Artificial Intelligence}, pubstate = {forthcoming}, tppubtype = {article} } We are witnessing a movement towards Democratizing AI, with a wide variety of tools, platforms and data sources becoming accessible to more people. This movement is said to be fueling innovation, extending the capabilities of individuals and organizations, by making the creation and application of AI easier. However, beyond the hype, there is a need to understand what this trend means for various stakeholders. Through the lens of socio-political democracy, this chapter examines the democratization of AI. We find that the present state of the “AI Democracy” maps onto only one of three elements of a democracy. Current efforts focus primarily on providing people with the tools and technical infrastructure needed to participate in AI, but not in protecting their freedoms and access to social benefits, which are the other core elements of democracy. We discuss the possibilities for realizing a broader AI democracy, along with the anticipated challenges. |
Kalia Orphanou Evgenia Christoforou, Jahna Otterbacher Monica Lestari Paramita Frank Hopfgartner Preserving the memory of the first wave of COVID-19 pandemic: Crowdsourcing a collection of image search queries Inproceedings 2021. Abstract | Links | BibTeX | Tags: Artificial Intelligence @inproceedings{Orphanou2021, title = {Preserving the memory of the first wave of COVID-19 pandemic: Crowdsourcing a collection of image search queries}, author = {Kalia Orphanou, Evgenia Christoforou, Jahna Otterbacher, Monica Lestari Paramita, Frank Hopfgartner}, url = {https://eprints.whiterose.ac.uk/180974/}, year = {2021}, date = {2021-11-10}, abstract = {The unprecedented events of the COVID-19 pandemic have generated an enormous amount of information and populated the Web with new content relevant to the pandemic and its implications. Visual information such as images has been shown to be crucial in the context of scientific communication. Images are often interpreted as being closer to the truth as compared to other forms of communication, because of their physical representation of an event such as the COVID-19 pandemic. In this work, we ask crowdworkers across four regions of Europe that were severely affected by the first wave of pandemic, to provide us with image search queries related to COVID-19 pandemic. The goal of this study is to understand the similarities/differences of the aspects that are most important to users across different locations regarding the first wave of COVID-19 pandemic. Through a content analysis of their queries, we discovered five common themes of concern to all, although the frequency of use differed across regions.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {inproceedings} } The unprecedented events of the COVID-19 pandemic have generated an enormous amount of information and populated the Web with new content relevant to the pandemic and its implications. Visual information such as images has been shown to be crucial in the context of scientific communication. Images are often interpreted as being closer to the truth as compared to other forms of communication, because of their physical representation of an event such as the COVID-19 pandemic. In this work, we ask crowdworkers across four regions of Europe that were severely affected by the first wave of pandemic, to provide us with image search queries related to COVID-19 pandemic. The goal of this study is to understand the similarities/differences of the aspects that are most important to users across different locations regarding the first wave of COVID-19 pandemic. Through a content analysis of their queries, we discovered five common themes of concern to all, although the frequency of use differed across regions. |
Styliani Kleanthous Maria Kasinidou, Pınar Barlas Jahna Otterbacher Perception of fairness in algorithmic decisions: Future developers' perspective Journal Article Patterns, 2021. Abstract | Links | BibTeX | Tags: Accountability, Algorithmic Fairness, Algorithmic Transparency, Artificial Intelligence @article{Kleanthous2021, title = {Perception of fairness in algorithmic decisions: Future developers' perspective}, author = {Styliani Kleanthous, Maria Kasinidou, Pınar Barlas, Jahna Otterbacher}, url = {https://www.sciencedirect.com/science/article/pii/S2666389921002476}, year = {2021}, date = {2021-11-03}, journal = {Patterns}, abstract = {Fairness, accountability, transparency, and ethics (FATE) in algorithmic systems is gaining a lot of attention lately. With the continuous advancement of machine learning and artificial intelligence, research and tech companies are coming across incidents where algorithmic systems are making non-objective decisions that may reproduce and/or amplify social stereotypes and inequalities. There is a great effort by the research community on developing frameworks of fairness and algorithmic models to alleviate biases; however, we first need to understand how people perceive the complex construct of algorithmic fairness. In this work, we investigate how young and future developers perceive these concepts. Our results can inform future research on (1) understanding perceptions of algorithmic FATE, (2) highlighting the needs for systematic training and education on FATE, and (3) raising awareness among young developers on the potential impact that the systems they are developing have in society.}, keywords = {Accountability, Algorithmic Fairness, Algorithmic Transparency, Artificial Intelligence}, pubstate = {published}, tppubtype = {article} } Fairness, accountability, transparency, and ethics (FATE) in algorithmic systems is gaining a lot of attention lately. With the continuous advancement of machine learning and artificial intelligence, research and tech companies are coming across incidents where algorithmic systems are making non-objective decisions that may reproduce and/or amplify social stereotypes and inequalities. There is a great effort by the research community on developing frameworks of fairness and algorithmic models to alleviate biases; however, we first need to understand how people perceive the complex construct of algorithmic fairness. In this work, we investigate how young and future developers perceive these concepts. Our results can inform future research on (1) understanding perceptions of algorithmic FATE, (2) highlighting the needs for systematic training and education on FATE, and (3) raising awareness among young developers on the potential impact that the systems they are developing have in society. |
Monica Lestari Paramita Kalia Orphanou, Evgenia Christoforou Jahna Otterbacher Frank Hopfgartner Do you see what I see? Images of the COVID-19 pandemic through the lens of Google Inproceedings 2021. Abstract | Links | BibTeX | Tags: Algorithmic Bias, Artificial Intelligence @inproceedings{Paramita2021, title = {Do you see what I see? Images of the COVID-19 pandemic through the lens of Google}, author = {Monica Lestari Paramita, Kalia Orphanou, Evgenia Christoforou, Jahna Otterbacher, Frank Hopfgartner}, url = {https://www.sciencedirect.com/science/article/pii/S0306457321001424}, year = {2021}, date = {2021-09-05}, journal = {Information Processing & Management}, abstract = {During times of crisis, information access is crucial. Given the opaque processes behind modern search engines, it is important to understand the extent to which the “picture” of the Covid-19 pandemic accessed by users differs. We explore variations in what users “see” concerning the pandemic through Google image search, using a two-step approach. First, we crowdsource a search task to users in four regions of Europe, asking them to help us create a photo documentary of Covid-19 by providing image search queries. Analysing the queries, we find five common themes describing information needs. Next, we study three sources of variation – users’ information needs, their geo-locations and query languages – and analyse their influences on the similarity of results. We find that users see the pandemic differently depending on where they live, as evidenced by the 46% similarity across results. When users expressed a given query in different languages, there was no overlap for most of the results. Our analysis suggests that localisation plays a major role in the (dis)similarity of results, and provides evidence of the diverse “picture” of the pandemic seen through Google.}, keywords = {Algorithmic Bias, Artificial Intelligence}, pubstate = {published}, tppubtype = {inproceedings} } During times of crisis, information access is crucial. Given the opaque processes behind modern search engines, it is important to understand the extent to which the “picture” of the Covid-19 pandemic accessed by users differs. We explore variations in what users “see” concerning the pandemic through Google image search, using a two-step approach. First, we crowdsource a search task to users in four regions of Europe, asking them to help us create a photo documentary of Covid-19 by providing image search queries. Analysing the queries, we find five common themes describing information needs. Next, we study three sources of variation – users’ information needs, their geo-locations and query languages – and analyse their influences on the similarity of results. We find that users see the pandemic differently depending on where they live, as evidenced by the 46% similarity across results. When users expressed a given query in different languages, there was no overlap for most of the results. Our analysis suggests that localisation plays a major role in the (dis)similarity of results, and provides evidence of the diverse “picture” of the pandemic seen through Google. |
Pınar Barlas Kyriakos Kyriakou, Styliani Kleanthous Jahna Otterbacher Person, Human, Neither: The Dehumanization Potential of Automated Image Tagging Proceeding 2021, ISBN: 9781450384735. Abstract | Links | BibTeX | Tags: Artificial Intelligence @proceedings{Barlas2021, title = {Person, Human, Neither: The Dehumanization Potential of Automated Image Tagging}, author = {Pınar Barlas, Kyriakos Kyriakou, Styliani Kleanthous, Jahna Otterbacher}, url = {https://dl.acm.org/doi/abs/10.1145/3461702.3462567}, doi = {10.1145/3461702.3462567}, isbn = {9781450384735}, year = {2021}, date = {2021-05-19}, series = {AIES '21}, abstract = {Following the literature on dehumanization via technology, we audit six proprietary image tagging algorithms (ITAs) for their potential to perpetuate dehumanization. We examine the ITAs' outputs on a controlled dataset of images depicting a diverse group of people for tags that indicate the presence of a human in the image. Through an analysis of the (mis)use of these tags, we find that there are some individuals whose 'humanness' is not recognized by an ITA, and that these individuals are often from marginalized social groups. Finally, we compare these findings with the use of the 'face' tag, which can be used for surveillance, revealing that people's faces are often recognized by an ITA even when their 'humanness' is not. Overall, we highlight the subtle ways in which ITAs may inflict widespread, disparate harm, and emphasize the importance of considering the social context of the resulting application.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {proceedings} } Following the literature on dehumanization via technology, we audit six proprietary image tagging algorithms (ITAs) for their potential to perpetuate dehumanization. We examine the ITAs' outputs on a controlled dataset of images depicting a diverse group of people for tags that indicate the presence of a human in the image. Through an analysis of the (mis)use of these tags, we find that there are some individuals whose 'humanness' is not recognized by an ITA, and that these individuals are often from marginalized social groups. Finally, we compare these findings with the use of the 'face' tag, which can be used for surveillance, revealing that people's faces are often recognized by an ITA even when their 'humanness' is not. Overall, we highlight the subtle ways in which ITAs may inflict widespread, disparate harm, and emphasize the importance of considering the social context of the resulting application. |
Kyriakos Kyriakou Pınar Barlas, Styliani Kleanthous Jahna Otterbacher OpenTag: Understanding Human Perceptions of Image Tagging Algorithms Conference HCOMP-20 2020. Abstract | Links | BibTeX | Tags: Artificial Intelligence @conference{Kyriakou2020b, title = {OpenTag: Understanding Human Perceptions of Image Tagging Algorithms}, author = {Kyriakos Kyriakou, Pınar Barlas, Styliani Kleanthous, Jahna Otterbacher}, url = {http://www.cycat.io/hcomp_2020_paper_76-2/}, year = {2020}, date = {2020-10-25}, series = {HCOMP-20}, abstract = {Image Tagging Algorithms (ITAs) are extensively used in our information ecosystem, from facilitating the retrieval of images in social platforms to learning about users and their preferences. However, audits performed on ITAs have demonstrated that their behaviors often exhibit social biases, especially when analyzing images depicting people. We present OpenTag, a platform that fuses the auditing process with a crowdsourcing approach. Users can upload an image, which is then analyzed by various ITAs, resulting in multiple sets of descriptive tags. With OpenTag, the user can observe and compare the output of multiple ITAs simultaneously, while researchers can study the manner in which users perceive this output. Finally, using the collected data, further audits can be performed on ITAs.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {conference} } Image Tagging Algorithms (ITAs) are extensively used in our information ecosystem, from facilitating the retrieval of images in social platforms to learning about users and their preferences. However, audits performed on ITAs have demonstrated that their behaviors often exhibit social biases, especially when analyzing images depicting people. We present OpenTag, a platform that fuses the auditing process with a crowdsourcing approach. Users can upload an image, which is then analyzed by various ITAs, resulting in multiple sets of descriptive tags. With OpenTag, the user can observe and compare the output of multiple ITAs simultaneously, while researchers can study the manner in which users perceive this output. Finally, using the collected data, further audits can be performed on ITAs. |
Evgenia Christoforou Pınar Barlas, Jahna Otterbacher Crowdwork as a Snapshot in Time: Image Annotation Tasks during a Pandemic Conference HCOMP-20 2020. Abstract | Links | BibTeX | Tags: Artificial Intelligence @conference{Christoforou2020, title = {Crowdwork as a Snapshot in Time: Image Annotation Tasks during a Pandemic}, author = {Evgenia Christoforou, Pınar Barlas, Jahna Otterbacher}, url = {http://www.cycat.io/hcomp_2020_paper_79/}, year = {2020}, date = {2020-10-25}, series = {HCOMP-20}, abstract = {While crowdsourcing provides a convenient solution for tapping into human intelligence, a concern is the bias inherent in the data collected. Events related to the COVID-19 pandemic had an impact on people globally, and crowdworkers were no exception. Given the evidence concerning mood and stress on work, we explore how temporal events might affect crowdsourced data. We replicated an image annotation task conducted in 2018, in which workers describe people images. We expected 2020 annotations to contain more references to health, as compared to 2018 data. Overall, we find no evidence that health-related tags were used more often in 2020, but instead we find a significant increase in the use of tags related to weight (e.g., fat, chubby, overweight). This result, coupled with the “stay at home” act in effect in 2020, illustrate how crowdwork is impacted by temporal events. }, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {conference} } While crowdsourcing provides a convenient solution for tapping into human intelligence, a concern is the bias inherent in the data collected. Events related to the COVID-19 pandemic had an impact on people globally, and crowdworkers were no exception. Given the evidence concerning mood and stress on work, we explore how temporal events might affect crowdsourced data. We replicated an image annotation task conducted in 2018, in which workers describe people images. We expected 2020 annotations to contain more references to health, as compared to 2018 data. Overall, we find no evidence that health-related tags were used more often in 2020, but instead we find a significant increase in the use of tags related to weight (e.g., fat, chubby, overweight). This result, coupled with the “stay at home” act in effect in 2020, illustrate how crowdwork is impacted by temporal events. |
Barlas, Pınar ; Kyriakou, Kyriakos ; Guest, Olivia ; Kleanthous, Styliani ; Otterbacher, Jahna 2020. Abstract | Links | BibTeX | Tags: Algorithmic Bias, Artificial Intelligence @proceedings{Barlas2020b, title = {To "See" is to Stereotype: Image Tagging Algorithms, Gender Recognition, and the Accuracy-Fairness Trade-off}, author = {Barlas, Pınar and Kyriakou, Kyriakos and Guest, Olivia and Kleanthous, Styliani and Otterbacher, Jahna}, url = {https://dl.acm.org/doi/abs/10.1145/3432931}, doi = {10.1145/3432931}, year = {2020}, date = {2020-10-17}, series = {CSCW3 20}, abstract = {Machine-learned computer vision algorithms for tagging images are increasingly used by developers and researchers, having become popularized as easy-to-use "cognitive services." Yet these tools struggle with gender recognition, particularly when processing images of women, people of color and non-binary individuals. Socio-technical researchers have cited data bias as a key problem; training datasets often over-represent images of people and contexts that convey social stereotypes. The social psychology literature explains that people learn social stereotypes, in part, by observing others in particular roles and contexts, and can inadvertently learn to associate gender with scenes, occupations and activities. Thus, we study the extent to which image tagging algorithms mimic this phenomenon. We design a controlled experiment, to examine the interdependence between algorithmic recognition of context and the depicted person's gender. In the spirit of auditing to understand machine behaviors, we create a highly controlled dataset of people images, imposed on gender-stereotyped backgrounds. Our methodology is reproducible and our code publicly available. Evaluating five proprietary algorithms, we find that in three, gender inference is hindered when a background is introduced. Of the two that "see" both backgrounds and gender, it is the one whose output is most consistent with human stereotyping processes that is superior in recognizing gender. We discuss the accuracy--fairness trade-off, as well as the importance of auditing black boxes in better understanding this double-edged sword.}, keywords = {Algorithmic Bias, Artificial Intelligence}, pubstate = {published}, tppubtype = {proceedings} } Machine-learned computer vision algorithms for tagging images are increasingly used by developers and researchers, having become popularized as easy-to-use "cognitive services." Yet these tools struggle with gender recognition, particularly when processing images of women, people of color and non-binary individuals. Socio-technical researchers have cited data bias as a key problem; training datasets often over-represent images of people and contexts that convey social stereotypes. The social psychology literature explains that people learn social stereotypes, in part, by observing others in particular roles and contexts, and can inadvertently learn to associate gender with scenes, occupations and activities. Thus, we study the extent to which image tagging algorithms mimic this phenomenon. We design a controlled experiment, to examine the interdependence between algorithmic recognition of context and the depicted person's gender. In the spirit of auditing to understand machine behaviors, we create a highly controlled dataset of people images, imposed on gender-stereotyped backgrounds. Our methodology is reproducible and our code publicly available. Evaluating five proprietary algorithms, we find that in three, gender inference is hindered when a background is introduced. Of the two that "see" both backgrounds and gender, it is the one whose output is most consistent with human stereotyping processes that is superior in recognizing gender. We discuss the accuracy--fairness trade-off, as well as the importance of auditing black boxes in better understanding this double-edged sword. |
Barlas, Pınar ; Kyriakou, Kyriakos ; Chrysanthou, Antrea ; Kleanthous, Styliani ; Otterbacher, Jahna OPIAS: Over-Personalization in Information Access Systems Inproceedings Adjunct Publication of the 28th ACM Conference on User Modeling, Adaptation and Personalization, pp. 103–104, 2020, ISBN: 9781450379502. Links | BibTeX | Tags: Artificial Intelligence @inproceedings{Barlas2020, title = {OPIAS: Over-Personalization in Information Access Systems}, author = {Barlas, Pınar and Kyriakou, Kyriakos and Chrysanthou, Antrea and Kleanthous, Styliani and Otterbacher, Jahna}, url = {https://dl.acm.org/doi/abs/10.1145/3386392.3397607}, doi = {10.1145/3386392.3397607}, isbn = {9781450379502}, year = {2020}, date = {2020-07-12}, booktitle = {Adjunct Publication of the 28th ACM Conference on User Modeling, Adaptation and Personalization}, pages = {103–104}, series = {UMAP '20 Adjunct}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {inproceedings} } |
Kyriakou, Kyriakos ; Kleanthous, Styliani ; Otterbacher, Jahna ; Papadopoulos, George A Emotion-based Stereotypes in Image Analysis Services Inproceedings Adjunct Publication of the 28th ACM Conference on User Modeling, Adaptation and Personalization, pp. 252–259, 2020, ISBN: 9781450379502. Abstract | Links | BibTeX | Tags: Artificial Intelligence @inproceedings{Kyriakou2020, title = {Emotion-based Stereotypes in Image Analysis Services}, author = {Kyriakou, Kyriakos and Kleanthous, Styliani and Otterbacher, Jahna and Papadopoulos, George A.}, url = {https://dl.acm.org/doi/abs/10.1145/3386392.3399567}, doi = {10.1145/3386392.3399567}, isbn = {9781450379502}, year = {2020}, date = {2020-07-12}, booktitle = {Adjunct Publication of the 28th ACM Conference on User Modeling, Adaptation and Personalization}, pages = {252–259}, series = {UMAP '20 Adjunct}, abstract = {Vision-based cognitive services (CogS) have become crucial in a wide range of applications, from real-time security and social networks to smartphone applications. Many services focus on analyzing people images. When it comes to facial analysis, these services can be misleading or even inaccurate, raising ethical concerns such as the amplification of social stereotypes. We analyzed popular Image Tagging CogS that infer emotion from a person's face, considering whether they perpetuate racial and gender stereotypes concerning emotion. By comparing both CogS and Human-generated descriptions on a set of controlled images, we highlight the need for transparency and fairness in CogS. In particular, we document evidence that CogS may actually be more likely than crowdworkers to perpetuate the stereotype of the "angry black man" and often attribute black race individuals with "emotions of hostility".}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {inproceedings} } Vision-based cognitive services (CogS) have become crucial in a wide range of applications, from real-time security and social networks to smartphone applications. Many services focus on analyzing people images. When it comes to facial analysis, these services can be misleading or even inaccurate, raising ethical concerns such as the amplification of social stereotypes. We analyzed popular Image Tagging CogS that infer emotion from a person's face, considering whether they perpetuate racial and gender stereotypes concerning emotion. By comparing both CogS and Human-generated descriptions on a set of controlled images, we highlight the need for transparency and fairness in CogS. In particular, we document evidence that CogS may actually be more likely than crowdworkers to perpetuate the stereotype of the "angry black man" and often attribute black race individuals with "emotions of hostility". |
Chrysanthou, Antrea ; Barlas, Pınar ; Kyriakou, Kyriakos ; Kleanthous, Styliani ; Otterbacher, Jahna Bursting the Bubble: Tool for Awareness and Research about Overpersonalization in Information Access Systems Inproceedings Proceedings of the 25th International Conference on Intelligent User Interfaces Companion, pp. 112–113, 2020, ISBN: 9781450375139. Abstract | Links | BibTeX | Tags: Artificial Intelligence @inproceedings{chrysanthou2020bursting, title = {Bursting the Bubble: Tool for Awareness and Research about Overpersonalization in Information Access Systems}, author = {Chrysanthou, Antrea and Barlas, Pınar and Kyriakou, Kyriakos and Kleanthous, Styliani and Otterbacher, Jahna}, url = {https://dl.acm.org/doi/abs/10.1145/3379336.3381863}, doi = {10.1145/3379336.3381863}, isbn = {9781450375139}, year = {2020}, date = {2020-03-17}, booktitle = {Proceedings of the 25th International Conference on Intelligent User Interfaces Companion}, pages = {112–113}, series = {IUI '20}, abstract = {Modern information access systems extensively use personalization, automatically filtering and/or ranking content based on the user profile, to guide users to the most relevant material. However, this can also lead to unwanted effects such as the "filter bubble." We present an interactive demonstration system, designed as an educational and research tool, which imitates a search engine, personalizing the search results returned for a query based on the user's characteristics. The system can be tailored to suit any type of audience and context, as well as enabling the collection of responses and interaction data.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {inproceedings} } Modern information access systems extensively use personalization, automatically filtering and/or ranking content based on the user profile, to guide users to the most relevant material. However, this can also lead to unwanted effects such as the "filter bubble." We present an interactive demonstration system, designed as an educational and research tool, which imitates a search engine, personalizing the search results returned for a query based on the user's characteristics. The system can be tailored to suit any type of audience and context, as well as enabling the collection of responses and interaction data. |
Otterbacher, Jahna ; Barlas, Pınar ; Kleanthous, Styliani ; Kyriakou, Kyriakos How Do We Talk about Other People? Group (Un)Fairness in Natural Language Image Descriptions Inproceedings Proceedings of the AAAI Conference on Human Computation and Crowdsourcing, pp. 106-114, 2019. Abstract | Links | BibTeX | Tags: Artificial Intelligence @inproceedings{otterbacher2019we, title = {How Do We Talk about Other People? Group (Un)Fairness in Natural Language Image Descriptions}, author = {Otterbacher, Jahna and Barlas, Pınar and Kleanthous, Styliani and Kyriakou, Kyriakos}, url = {https://ojs.aaai.org/index.php/HCOMP/article/view/5267}, year = {2019}, date = {2019-10-28}, booktitle = {Proceedings of the AAAI Conference on Human Computation and Crowdsourcing}, volume = {7}, number = {1}, pages = {106-114}, series = {HCOMP-19}, abstract = {Crowdsourcing plays a key role in developing algorithms for image recognition or captioning. Major datasets, such as MS COCO or Flickr30K, have been built by eliciting natural language descriptions of images from workers. Yet such elicitation tasks are susceptible to human biases, including stereotyping people depicted in images. Given the growing concerns surrounding discrimination in algorithms, as well as in the data used to train them, it is necessary to take a critical look at this practice. We conduct experiments at Figure Eight using a controlled set of people images. Men and women of various races are positioned in the same manner, wearing a grey t-shirt. We prompt workers for 10 descriptive labels, and consider them using the human-centric approach, which assumes reporting bias. We find that “what’s worth saying” about these uniform images often differs as a function of the gender and race of the depicted person, violating the notion of group fairness. Although this diversity in natural language people descriptions is expected and often beneficial, it could result in automated disparate impact if not managed properly.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {inproceedings} } Crowdsourcing plays a key role in developing algorithms for image recognition or captioning. Major datasets, such as MS COCO or Flickr30K, have been built by eliciting natural language descriptions of images from workers. Yet such elicitation tasks are susceptible to human biases, including stereotyping people depicted in images. Given the growing concerns surrounding discrimination in algorithms, as well as in the data used to train them, it is necessary to take a critical look at this practice. We conduct experiments at Figure Eight using a controlled set of people images. Men and women of various races are positioned in the same manner, wearing a grey t-shirt. We prompt workers for 10 descriptive labels, and consider them using the human-centric approach, which assumes reporting bias. We find that “what’s worth saying” about these uniform images often differs as a function of the gender and race of the depicted person, violating the notion of group fairness. Although this diversity in natural language people descriptions is expected and often beneficial, it could result in automated disparate impact if not managed properly. |
Batsuren Khuyagbaatar Ganbold Amarsanaa, Chagnaa Altangerel Giunchiglia Fausto Building the mongolian wordnet Inproceedings Proceedings of the 10th global WordNet conference, 2019. Abstract | Links | BibTeX | Tags: Artificial Intelligence @inproceedings{Khuyagbaatar2019, title = {Building the mongolian wordnet}, author = {Batsuren Khuyagbaatar, Ganbold Amarsanaa, Chagnaa Altangerel, Giunchiglia Fausto}, url = {https://aclanthology.org/2019.gwc-1.30}, year = {2019}, date = {2019-07-08}, booktitle = {Proceedings of the 10th global WordNet conference}, abstract = {This paper presents the Mongolian Wordnet (MOW), and a general methodology of how to construct it from various sources e.g. lexical resources and expert translations. As of today, the MOW contains 23,665 synsets, 26,875 words, 2,979 glosses, and 213 examples. The manual evaluation of the resource estimated its quality at 96.4%.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {inproceedings} } This paper presents the Mongolian Wordnet (MOW), and a general methodology of how to construct it from various sources e.g. lexical resources and expert translations. As of today, the MOW contains 23,665 synsets, 26,875 words, 2,979 glosses, and 213 examples. The manual evaluation of the resource estimated its quality at 96.4%. |
Kyriakou, Kyriakos; Barlas, Pınar; Kleanthous, Styliani; Otterbacher, Jahna Fairness in Proprietary Image Tagging Algorithms: A Cross-Platform Audit on People Images Conference ICWSM 2019 AAAI, 2019, ISSN: 2334-0770. Abstract | Links | BibTeX | Tags: Artificial Intelligence @conference{KyriakouICWSM2019, title = {Fairness in Proprietary Image Tagging Algorithms: A Cross-Platform Audit on People Images}, author = {Kyriakos Kyriakou and Pınar Barlas and Styliani Kleanthous and Jahna Otterbacher}, url = {http://www.cycat.io/wp-content/uploads/2019/05/ICWSM_tagging_b_eye_as_v4-2.pdf}, issn = {2334-0770}, year = {2019}, date = {2019-06-15}, publisher = {AAAI}, series = {ICWSM 2019}, abstract = {There are increasing expectations that algorithms should be- have in a manner that is socially just. We consider the case of image tagging APIs and their interpretations of people im- ages. Image taggers have become indispensable in our in- formation ecosystem, facilitating new modes of visual com- munication and sharing. Recently, they have become widely available as Cognitive Services. But while tagging APIs of- fer developers an inexpensive and convenient means to add functionality to their creations, most are opaque and propri- etary. Through a cross-platform comparison of six taggers, we show that behaviors differ significantly. While some of- fer more interpretation on images, they may exhibit less fair- ness toward the depicted persons, by misuse of gender-related tags and/or making judgments on a person’s physical appear- ance. We also discuss the difficulties of studying fairness in situations where algorithmic systems cannot be benchmarked against a ground truth.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {conference} } There are increasing expectations that algorithms should be- have in a manner that is socially just. We consider the case of image tagging APIs and their interpretations of people im- ages. Image taggers have become indispensable in our in- formation ecosystem, facilitating new modes of visual com- munication and sharing. Recently, they have become widely available as Cognitive Services. But while tagging APIs of- fer developers an inexpensive and convenient means to add functionality to their creations, most are opaque and propri- etary. Through a cross-platform comparison of six taggers, we show that behaviors differ significantly. While some of- fer more interpretation on images, they may exhibit less fair- ness toward the depicted persons, by misuse of gender-related tags and/or making judgments on a person’s physical appear- ance. We also discuss the difficulties of studying fairness in situations where algorithmic systems cannot be benchmarked against a ground truth. |
Barlas, Pınar; Kyriakou, Kyriakos; Kleanthous, Styliani; Otterbacher, Jahna Social B(eye)as: Human and Machine Descriptions of People Images Conference ICWSM 2019 AAAI, 2019, ISSN: 2334-0770. Abstract | Links | BibTeX | Tags: Artificial Intelligence @conference{BarlasICWSM2019, title = {Social B(eye)as: Human and Machine Descriptions of People Images}, author = {Pınar Barlas and Kyriakos Kyriakou and Styliani Kleanthous and Jahna Otterbacher}, url = {http://www.cycat.io/wp-content/uploads/2019/05/ICWSM_dataset_CAMERAREADY-2.pdf}, issn = {2334-0770}, year = {2019}, date = {2019-06-15}, publisher = {AAAI}, series = {ICWSM 2019}, abstract = {Image analysis algorithms have become an indispensable tool in our information ecosystem, facilitating new forms of visual communication and information sharing. At the same time, they enable large-scale socio-technical research which would otherwise be difficult to carry out. However, their outputs may exhibit social bias, especially when analyzing people images. Since most algorithms are proprietary and opaque, we propose a method of auditing their outputs for social biases. To be able to compare how algorithms interpret a controlled set of people images, we collected descriptions across six image tagging algorithms. In order to compare these results to human behavior, we also collected descriptions on the same images from crowdworkers in two anglophone regions. The dataset we present consists of tags from these eight taggers, along with a typology of concepts, and a python script to calculate vector scores for each image and tagger. Using our methodology, researchers can see the behaviors of the image tagging algorithms and compare them to those of crowdworkers. Beyond computer vision auditing, the dataset of human- and machine-produced tags, the typology, and the vectorization method can be used to explore a range of research questions related to both algorithmic and human behaviors.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {conference} } Image analysis algorithms have become an indispensable tool in our information ecosystem, facilitating new forms of visual communication and information sharing. At the same time, they enable large-scale socio-technical research which would otherwise be difficult to carry out. However, their outputs may exhibit social bias, especially when analyzing people images. Since most algorithms are proprietary and opaque, we propose a method of auditing their outputs for social biases. To be able to compare how algorithms interpret a controlled set of people images, we collected descriptions across six image tagging algorithms. In order to compare these results to human behavior, we also collected descriptions on the same images from crowdworkers in two anglophone regions. The dataset we present consists of tags from these eight taggers, along with a typology of concepts, and a python script to calculate vector scores for each image and tagger. Using our methodology, researchers can see the behaviors of the image tagging algorithms and compare them to those of crowdworkers. Beyond computer vision auditing, the dataset of human- and machine-produced tags, the typology, and the vectorization method can be used to explore a range of research questions related to both algorithmic and human behaviors. |
Barlas, Pınar; Kyriakou, Kyriakos; Kleanthous, Styliani; Otterbacher, Jahna What Makes an Image Tagger Fair? - Proprietary Auto-tagging and Interpretations on People Images Conference UMAP 2019 ACM, 2019. Abstract | Links | BibTeX | Tags: Artificial Intelligence @conference{BarlasUMAP2019, title = {What Makes an Image Tagger Fair? - Proprietary Auto-tagging and Interpretations on People Images}, author = {Pınar Barlas and Kyriakos Kyriakou and Styliani Kleanthous and Jahna Otterbacher}, url = {http://www.cycat.io/wp-content/uploads/2019/05/Barlas-et-al.-2019-What-Makes-an-Image-Tagger-Fair-Proprietary-Auto-tagging-and-Interpretations-on-People-Images-1.pdf}, doi = {10.1145/3320435.3320442}, year = {2019}, date = {2019-06-13}, publisher = {ACM}, series = {UMAP 2019}, abstract = {Image analysis algorithms have been a boon to personalization in digital systems and are now widely available via easy-to-use APIs. However, it is important to ensure that they behave fairly in applications that involve processing images of people, such as dating apps. We conduct an experiment to shed light on the factors influencing the perception of “fairness." Participants are shown a photo along with two descriptions (human- and algorithm-generated). They are then asked to indicate which is “more fair" in the context of a dating site, and explain their reasoning. We vary a number of factors, including the gender, race and attractiveness of the person in the photo. While participants generally found human-generated tags to be more fair, API tags were judged as being more fair in one setting - where the image depicted an “attractive," white individual. In their explanations, participants often mention accuracy, as well as the objectivity/subjectivity of the tags in the description. We relate our work to the ongoing conversation about fairness in opaque tools like image tagging APIs, and their potential to result in harm.}, keywords = {Artificial Intelligence}, pubstate = {published}, tppubtype = {conference} } Image analysis algorithms have been a boon to personalization in digital systems and are now widely available via easy-to-use APIs. However, it is important to ensure that they behave fairly in applications that involve processing images of people, such as dating apps. We conduct an experiment to shed light on the factors influencing the perception of “fairness." Participants are shown a photo along with two descriptions (human- and algorithm-generated). They are then asked to indicate which is “more fair" in the context of a dating site, and explain their reasoning. We vary a number of factors, including the gender, race and attractiveness of the person in the photo. While participants generally found human-generated tags to be more fair, API tags were judged as being more fair in one setting - where the image depicted an “attractive," white individual. In their explanations, participants often mention accuracy, as well as the objectivity/subjectivity of the tags in the description. We relate our work to the ongoing conversation about fairness in opaque tools like image tagging APIs, and their potential to result in harm. |