Join Us

Open positions

Postdoc

No open positions

PhD

No open positions

Research assistant

No open positions

Master’s theses

Individual fairness in social network analysis

Description: Individual fairness dictates that similar individuals should have similar outcomes. This constraint is centered around the notion of a task-specific similarity metric describing the extent to which pairs of individuals should be regarded as similar for the task at hand. The goal of this project is to examine how the structural bias of social networks impacts the bias of social network analysis algorithms, under various similarity metrics and network topologies. More information
Requirements: Mining of Social Data
Contact: Theodora Moldovan

Anatomy of TikTok Videos

Description: Short-videos and recommender algorithms have become the dominant format within social media. While many studies started by mainly focusing on one or two of the modalities of the videos for their analysis, first studies have started to develop tools and frameworks to comprehend the multimodal aspects and how they shape the content, e.g. such as through framing. However, the temporal aspect of the videos is still understudied, despite it being of high relevance, as many users will only encounter the first (few) second(s) of the video, before scrolling to the next. This project aims to develop a pipeline to study the different modalities (audio, visuals, text) of a TikTok video across the temporal axis and apply it to a large-scale datasets. Of particular interest is the analysis of the mechanisms through which certain modalities might attract users’ attention or elicit an emotional response from them.

References:
[1] Zha, M., & Chang, H.-C. H. (2026). Interpreting multimodal communication at scale in short-form video: Visual, audio, and textual mental health discourse on TikTok. doi:10.48550/arXiv.2601.15278
[2] Hu, H. L., & Fu, K.-W. (2026). Seeing, reading, hearing, interpreting: Computational multimodal framing analysis in short-form news videos. Computational Communication Research, 8(4), 1–35. doi:10.5117/ccr2026.4.4.hu
Requirements: Data Mining & Social Data Mining, Advanced Probabilistic Machine Learning (meriting), Introduction to Image Analysis (meriting)
Contact: Inga Wohlert

Measuring Visual Political Polarization

Description: Visual content is central to contemporary online political communication, yet computational social science lacks well-developed measures of polarization in large-scale visual data. Recent work has emphasized the importance of treating images directly as political data and developed methods for identifying latent visual structure and frames at scale [1,2]. Building on broader approaches that conceptualize polarization as combining substantive distance with relational structure [3], this project aims to develop and validate computational measures of polarization in multimodal political communication.

The project distinguishes three related layers:

Visual repertoire polarization: How strongly do political groups occupy different regions of visual space? A key conceptual and methodological question is resolution: groups may share broad visual themes while diverging sharply in the specific scenes, actors, symbols, or motifs they select.

Interpretive polarization: Conditional on using the same or visually similar material, how differently do groups contextualize it through endorsement, criticism, irony, mockery, or reframing?

Relational polarization: How are visual repertoires and interpretations embedded in interaction networks, and when does shared content bridge groups versus become an object of antagonistic recontextualization?

Methodologically, the project will develop and validate computational measures combining visual representations, multiscale clustering or graph structure, contextual information, and network data. The broader goal is to measure how visual difference, interpretive difference, and relational separation jointly structure political polarization.

References:
[1] Williams, N. W., Casas, A., & Wilkerson, J. D. (2020). Images as Data for Social Science Research: An Introduction to Convolutional Neural Nets for Image Classification. Cambridge University Press.
[2] Torres, M. (2024). A framework for the unsupervised and semi-supervised analysis of visual frames. Political Analysis.
[3] Hohmann, M., Devriendt, K., & Coscia, M. (2023). Quantifying ideological polarization on a network using generalized Euclidean distance. Science Advances, 9(9).
Requirements: Mining of Social Data, Advanced Probabilistic Machine Learning (meriting), Introduction to Image Analysis (meriting), Advanced Deep Learning for Image Processing (meriting)
Contact: Matias Piqueras

Negotiating the Floor: Participation and Authority in Podcast Conversations

Description: This project investigates how podcast hosts and guests negotiate participation and conversational authority. Combining transcript analysis with audio processing, you will examine how participants gain, maintain, and yield the conversational floor, and how interruptions, brief vocal responses, laughter, pauses, and overlapping speech shape a shared narrative. The social aim is to understand whether guests have space to develop independent accounts, primarily reinforce the host’s interpretation, or challenge the direction of the conversation. Students will align transcripts and audio, annotate selected interactional events, and develop models of their temporal relationships. Comparing transcript-based models with models incorporating timing and acoustic features will help evaluate what audio contributes to understanding these interactions. Comparisons may examine different podcast formats, formal versus enacted host–guest roles, and supportive versus competitive participation, while accounting for recording and editing differences. More information
Requirements: Mining of Social Data, Deep Learning
Contact: Davide Vega

Clustering Conversations Using Hypergraphs and Stochastic Block Models

Description: Long conversations involve complex interactions between speakers, people directly addressed, and other participants who listen or respond. This project investigates how hypergraphs and stochastic block models (SBMs) can capture these relationships and reveal the structure of conversations in movies and podcasts. Students will represent each utterance as a hyperedge connecting its speaker, direct addressees, and indirect participants while preserving their different roles. Topical, prosodic, and temporal information will enrich this representation. The central task is to implement or adapt a hypergraph SBM and investigate whether the resulting clusters capture stable groups of participants, shared topics, or temporary patterns of cooperation and confrontation. Evaluation will use synthetic data, annotated scenes or conversational episodes, and comparisons with simpler clustering approaches and pairwise networks. The project requires a strong background in probabilistic inference and itnerest in algorithms and data structures. More information
Requirements: Mining of Social Data
Contact: Davide Vega

Comparing Election Interference Reporting Using Modern NLP Methods

Description: Debates about election interference concern both attempts to manipulate electoral outcomes and the ways allegations are used in political discourse. Drawing on framing analysis, this project investigates how public discussion of election interference develops across countries and over time. Students will collect and analyse their own corpus, drawing on sources such as newspapers or parliamentary speeches, and combine classical text analysis with state-of-the-art NLP methods. Tasks will include topic modelling to identify issues, frequency analysis to measure their salience, sentiment and framing analysis to examine how interference is portrayed, and temporal modelling to trace shifts in attention. Comparisons may explore established and newer democracies, languages, media systems, or communication platforms. The project connects data science with social communication research to understand how election interference is discussed and how allegations contribute to broader narratives about elections. No prior knowledge of Swedish language or culture is required; additional language skills may support cross-country comparisons. More information
Requirements: Mining of Social Data
Contact: Davide Vega

Operationalizing Social Theories for Multi-Agent AI Interaction in AI Village

Description: This thesis will study longitudinal interaction data from the AI Village, an environment where different AI agents communicate, use tools and pursue shared goals. The central methodological challenge is to translate concepts from social science into measurable indicators of agent behaviour. The project will begin with a data audit and then select one primary theoretical perspective, such as routine dynamics, institutionalization, coordination theory or social influence. Abstract constructs will be operationalized into measurable features using suitable computational methods. The analysis may combine NLP/text mining, temporal network analysis, sequence/event analysis and other methods. Potential questions include how repeated practices become stable routines, whether some agents acquire persistent influence or coordination roles, how conventions spread across the group, and how agents respond when shared expectations are violated. The goal is not to claim that AI agents possess human social psychology, but to investigate whether concepts from social theory provide useful tools for describing and explaining observable collective behaviour in multi-agent AI systems.
Requirements: Mining of Social Data
Contact: Ece Calikus

Motif-Based Identification of the ECG Features Associated with Raised Intracranial Pressure

Description: Traumatic brain injury (TBI) is one of the most important causes of death and disability worldwide, with an estimated 27 million new cases each year. Guidelines for TBI care emphasise that monitoring and treatment should be guided by increased intracranial pressure (ICP), but reliable ICP monitoring requires a neurosurgical procedure to place a sensor in the ventricles of the brain, making it unavailable to the majority of patients globally. Continuous electrocardiogram (ECG) is non-invasive and widely used for cardiac monitoring. ECG changes in severe neurological insult are clinically well recognised and are thought to arise from increased sympathetic activation and autonomic dysregulation secondary to intracranial pathology. A completed pilot analysis in this project (88 patients, 4.5 million 10-second windows) found that the ECG beat departs progressively further from its baseline shape as ICP increases, suggesting the association may be dose-dependent. This project addresses the explanatory question: which ECG features account for the association between the ECG and ICP? The project uses the MIMIC-III Waveform Database (patients with concurrent invasive ICP monitoring and continuous single-lead ECG). All data are de-identified and openly available through PhysioNet, and the preprocessing pipeline (decoding to physical units, 10-second windowing, validity labelling, mean ICP per window) is already built.
Requirements: Machine Learning, Deep Learning
Contact: Ece Calikus

Bachelor theses

No open positions

Before joining

This section collects some information about life in Sweden and about how we work, that can be useful if you are interested in joining the lab as a PhD student or postdoc.

Working and living in Sweden

Sweden is a fantastic place for living and working. Swedes are friendly and speak excellent English. The quality of life is high, with a strong emphasis on outdoor activities. The Swedish working climate emphasises an open atmosphere, with active discussions involving both junior and senior staff. Spouses of employees are entitled to work permits. Healthcare is free after a small co-pay and the university subsidises athletic costs, such as a gym membership. The parental benefits in Sweden are among the best in the world, including extensive parental leave (for both parents), paid time off to care for sick children, and affordable daycare. For more information, be sure to read Why choose Sweden? and Why choose Uppsala University?.

Expectations from members of the InfoLab

Being part of a lab comes with many advantages, for example in terms of access to knowledge and information, but requires everyone to make a (small) effort to contribute to the lab environment. The following are some guidelines for the InfoLab.

Expected availability

Given the creative nature of part of our work, Infolab members are generally free to work when and where it best fits them, compatibly with tasks requiring presence (e.g., teaching) and regulations that may be enforced by the Department or University.

At the same time, everyone is expected to contribute to the lab, which implies being available to give feedback, sharing their current research with other members (see below), and participating in lab meetings. While respecting other people’s right to focus on their work, InfoLab members should feel welcome to knock on each other’s doors to ask questions and discuss work-related issues. If one is working from home one day, it is good practice to inform the others, because someone else might also have planned to work from home on the same day, but decided to go to the office specifically because they thought they could talk to you about something without needing to set up a formal meeting.

Lab members are free to take time to work offline. In particular, it is not necessary to be continuously online on our Slack space. Checking the Slack at least once a day is however expected, and if one cannot be online for a few days for any reason (e.g. holidays) it is good practice to inform the other lab members.

The information above concerns our lab. However, you are not employeed by the lab but by the division, and your contract is probably one expecting that you work at your office. Currently, our head of division is normally ok with us working at home some days, as long as this does not prevent us from performing our duties (which include being available if people need to talk to us). Therefore, if you plan to regularly work part of the time from home make sure that your immediate manager approves that.

Research quality at the Infolab

While “quality” is a subjective term, and we expect lab members to reflect about this concept and adopt their own definitions of quality, we also have some guidelines to ensure that some common ground is established. In particular, we want our research to be based on appropriate methods, clearly and transparently communicated, developed with ethical and societal awareness, replicable (when possible), and usable by others.

To achieve this, research at Infolab is organised into research activities. A research activity is anything leading to a published article, a grant proposal, a PhD dissertation, an undergraduate thesis, research software, etc. Each research activity is led by a member of the lab, who is the contact author and is responsible for the research to advance and for its quality. Under normal circumstances, the contact author should be available to answer questions about the activity and its outcomes for at least two years after the outcome has been made public, even in case of a change of affiliation.

Members of the lab should feel a responsibility of getting and receiving feedback from other members. At the beginning of a new research activity, the (contact) author(s) is encouraged to pitch the idea to other lab members, who should be available to provide constructive feedback. We trust each other not to share this information outside the lab, and we have no expectations that our feedback will be acknowledged in the resulting publications - we just do this to help each other.

At submission time, for all products containing experiments the code to replicate the experiments should be made available in a repository, in an easy-to-execute format (with exceptions, for example double-blind review processes).

Authorship

Each research activity must have a clear list of authors. At the Infolab we define authors as in: “each author is expected to have made substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data; or the creation of new software used in the work; or have drafted the work or substantively revised it; AND to have approved the submitted version (and any substantially modified version that involves the author’s contribution to the study); AND to have agreed both to be personally accountable for the author’s own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and the resolution documented in the literature.”¹

PhD supervision policy

Courses

In our PhD programme, every PhD student must obtain 60 credits from “courses”. While sometimes called “courses”, these can be very different from Master-level courses (which, in Uppsala, typically consist of multiple weekly lectures, labs, assignments, etc. over around 10 or 20 weeks). In some cases, a PhD course is just about individual readings and a written or oral presentation of the studied content. In other cases, courses involve lectures and interactions with other students, with either intensive meetings over a few consecutive days or less frequent meetings spread along the academic year. Some of the courses can however be actual Master’s courses, upon agreement with the supervisors and the course responsible teacher.

Courses are a fundamental part of the education, which (as indicated in the ISP goals) is not just about writing research papers but also about acquiring broad knowledge beyond the specific topic of the PhD and other soft and hard skills. “Research introduction for PhD students” is strongly recommended. “Research ethics” (at least 2 credits) and “Academic teacher training course” (7.5 credits, for those involved in teaching which is the default except for industrial PhD students) are mandatory and given by the Faculty. Some courses are given by InfoLab teachers and provide basic knowledge and skills in the general research field of the lab. Our students are expected to take these courses if they do not already have the corresponding knowledge or other pressing tasks to perform. Examples of courses given by Infolab members that are in the general research field of the lab can be found on our webpage.

Courses must always be agreed in advance with the main supervisor, as this is part of the PhD planning and because the number of credits that can be registered and used for the PhD is typically not pre-defined. For example, when reading a Master’s course: (1) as credits correspond to study time, it can sometimes be assumed that a PhD student will use less than the expected study time of an average Master’s student. (2) Some parts of the course may not be relevant for the PhD education. Some parts of a course can even be skipped, or replaced with more useful tasks, such as a custom project.

While your supervisor typically decides how many credits to assign to each course, these are general guidelines:

• For PhD-level courses given at our Faculty and open to all PhD students at the Faculty, e.g. teacher training, the number of credits is normally the one proposed by the teacher of the course.

• You may, in special cases, get credits if you teach MSc courses that you had not previously studied. The part of your employment devoted to teaching assistantships (TA) is paid by State funding for UGA (undergraduate education). Normally, part of your TA time is already allocated for you to prepare. If you teach a course that you have not taken in your previous studies, and the time allocated for preparation is not sufficient, your supervisors can decide to assign [alpha] x [credits of the theory part of the course] credits. The rationale is that at the end of the course you will have also learned part of the course content as if you had read the course. Alpha is decided by the supervisor based on content and duties for each course. No specific examination is needed to obtain these credits beyond being a TA.

• PhD students are welcome and encouraged to propose reading courses that can be of common interest for people in the lab and/or other PhD students at the Department. Feel free to discuss this with your fellow PhD students, and make a proposal to your supervisor.

• Credits for summer schools typically correspond to 1.5 for a week-long school, plus additional credits if additional work is needed for the examination. It happens that external summer schools promise unreasonable amounts of credits, so remember that the actual credits are decided by your supervisor.

• All research seminars given at the department or associated with the InfoLab (including PhD half-time seminars and the Lab’s lunchtime seminars) can be included in the ISP and can be later registered as credits. 40 hours of seminars correspond to 1.5 credits, with steps of 0.5 credits. Other seminars can be credited upon approval by the main supervisor.

• Conferences are considered equivalent to PhD schools, corresponding to 1.5 credits for a one-week conference where the student attends all the sessions. To obtain the credits, the student must present a summary of or highlights from the conference at a Lab research seminar.

Funding sources

Every PhD student is funded by a combination of different sources. Most of the time, this does not have an immediate impact on the student’s activities, and it is mainly a matter of interest for the supervisor. However, there are some obligations and opportunities associated with funding sources. These are the most typical:

• For externally funded projects, there is always a (more or less constraining) project proposal, sometimes in the form of formal contracts between the Department and the funding agency. Students funded by external projects can be expected to deliver specific (types of) results. Your supervisor is typically in charge of the funding (although it is always the Department being financially responsible and having to approve all expenses).

• For (partly) internally funded students, the Professor responsible for the Research Programme (PAP) is in charge of the funding. So any expenses beyond the salary have to be approved by the PAP. See the details about traveling below.

Travelling

For internally funded students, you have to follow the Division’s policy for travelling (which is the same for all research personnel, from PhD students to professors). While the details can be found in the original document and may change in time, a practical summary is that:

• Travelling expenses are normally covered if you travel to a conference to present your research.

• For PhD students, “normally” can be interpreted as: “the presentation must be important for the development of the PhD plan”. Your supervisor will make sure that this is the case when possible publication venues are discussed.

• In any case, a budget must be approved by the person responsible for the funding before the trip is booked.

• It is important to use these resources wisely, e.g. by looking for modest accommodation and travelling. This is (in most cases) money coming from tax-payers, taken from the same (limited) source used to fund the salary of other PhD students, etc.

The following guidelines complement the ones at division level:

• A typical (although in no way constraining) case is that each PhD student presents one work at a conference every year. This is important to train presentation skills and to build your network. For conferences without proceedings, the work is then submitted as a journal article. During the first year, when research results may not yet be ready for submission, a summer school in the area of the PhD can be considered as an alternative to a conference participation.

• When possible, climate-friendly travelling alternatives should be preferred. Given that conferences travel around the world and each major conference will be in Europe at least once and normally multiple times during a PhD education, when we submit our papers we try to favour conferences planned in Europe. We will still be able to be present at the important venues in our area, while also having a low environmental impact.

• During the PhD it can be good to participate in a PhD school (e.g. a summer school), as mentioned above. This can be done at the beginning, to get in touch with other relevant PhD students in your area, or later, to deepen your knowledge and network on a specific advanced topic that is important for your studies. The relevance of the school for your studies has to be argued in advance with the PAP (if to be covered with State funding), but participation in a PhD school during your studies is encouraged and will be supported by your supervisor.

• The fact that relevant and important travelling is generally covered by the Department means that you do not have to worry about this. At the same time, we expect our students to apply for external funding. This is part of your education (writing grant requests), it is good for your CV, it is common for our PhD students to indeed obtain funding, and this is funding you are in charge of, so in principle you can participate to as many conferences/schools/etc. as long as you can acquire those resources.

During the PhD studies, we expect all students to spend some time (normally a few months) at a different research centre, usually abroad. This is important for many reasons - including experiencing different ways of doing research, different supervision styles, networking, access to resources, etc. You keep receiving your salary during this stay, but often there are additional expenses (e.g. having to rent a new apartment while you may not be able to rent out yours in Uppsala), so it is particularly important to look for external funding. Various sources are available.

Papers and co-authorship

When you start working on a new research task as part of your PhD, you must always involve your supervisor(s). This is not intended as a way to control you - in fact, at the end of your PhD you should be able to design and write a paper without the need for your supervisor’s input. However, especially at the beginning of your studies there is a risk that you are missing some information or questions that you have not thought about (e.g. You are using the Department’s and lab affiliations - what does this imply? Is this research topic compatible with the sources funding it?). This is also the case for side-research-projects that are intended to be done in your spare time - as these, by experience, have a high chance of interacting with your PhD work, and should thus be planned jointly.

When you write papers related to your PhD, your supervisors will also typically be your co-authors. At the lab we follow international guidelines about co-authorship (see above), but ultimately the rules and meaning of being a co-author is something discipline-specific. Even explicitly specifying the contribution of each author in the paper (which is recommended in general) can send different messages depending on the discipline. For example, in most areas of computer science your supervisor is expected to be specified, and an additional note specifying that the contribution of the person has been “supervision” can be interpreted as suggesting a lower involvement than one would otherwise assume. Not having your supervisor as a co-author might even suggest to future prospective employers that there has been some conflict. On the contrary, in other areas, having your supervisor as a co-author is not expected, and would raise questions about your role and independence. This is an important and challenging topic for our lab, being part of several interdisciplinary collaborations. The default behaviour is that (1) your supervisor(s) who meet our general co-authorship criteria will appear as co-authors; (2) for venues where this is expected or useful, the contributions of each co-author will be explicitly mentioned in the paper, and (3) individual special cases must always be discussed in advance.

1. Transparency in authors’ contributions and responsibilities to promote integrity in scientific publication. Marcia K. McNutt, Monica Bradford, Jeffrey M. Drazen, Brooks Hanson, BobHoward, Kathleen Hall Jamieson, Véronique Kiermer, Emilie Marcus, Barbara Kline Pope, Randy Schekman, Sowmya Swaminathan, Peter J. Stang, Inder M. Verma. Proceedings of the National Academy of Sciences Mar 2018, 115 (11) 2557-2560; DOI:10.1073/pnas.1715374115 ↩