WHAT IT IS
The Digital Practices Matrix (DPM) is an analytical tool developed within the research project Borrowing Algorithmic Epistemologies, coordinated by Prof. Elias Bitencourt at Universidade do Estado da Bahia, in partnership with Prof. Leonardo Pastor from Universidade Federal de Sergipe
DPM was created to explore digital practices—the ways groups and individuals use and appropriate platform grammars to communicate or consume content.
Inspired by recommendation systems and information retrieval techniques, the tool adapts these computational approaches to identify usage patterns, exclusivity, and significance of practices within different groups.
Beyond mapping what is communicated, DPM analyzes how practices structure communication and visibility, offering a resource to understand the social and cultural dynamics mediated by digital platforms.
Mapping the Dynamics of Digital Communication
PROBLEM AND MOTIVATION
Digital platforms like Instagram and TikTok condition interactions and visibility based on the use of interface functionalities, such as tagging, hashtags, emojis, posting times, media types, and other properties. These elements are mediated by algorithmic systems that use mathematical models to optimize recommendations and modulate visibility and reach based on recorded patterns.
Traditional methodologies, such as content analysis and topic modeling (LDA), or deep learning-based language models (like BERT), prioritize textual content and often overlook broader patterns of interface interaction. On the other hand, network analysis approaches often focus on relational structures between actors, frequently disregarding usage practices and interface appropriations essential for understanding communication dynamics.
The Digital Practices Matrix was developed to fill this gap, appropriating analytical techniques widely used in algorithmically mediated systems—which organize and produce knowledge from interaction patterns—to investigate how platforms and their algorithms influence and co-produce communication practices.
The tool seeks to explore not only what is communicated but how platforms are used to mediate visibility and engagement. This approach allows contextualizing the analyzed content by mapping how practices may affect online communication and visibility.
Digital Practices: Looking Beyond Content
CONTEXT
DPM was developed within the Borrowing Algorithmic Epistemologies research project, coordinated by Prof. Elias Bitencourt. The idea of this research is to explore techniques used in algorithmic systems to study digital practices. The analyses and conceptual developments were elaborated in partnership with Prof. Leonardo Pastor (UFS).
For the Digital Practices Matrix, we focused on the experimental use of term weighting and information retrieval techniques. In computational contexts, these techniques are combined with language models and machine learning to identify relevance, prioritize visibility, and organize content, as seen in recommendation systems.
Here, these techniques are reconfigured to map digital grammars and practices, offering a critical lens to understand how platform elements are appropriated in diverse ways.
This recontextualization does not treat the technique as a black box but as a resource to explore digital interactions and practices. By doing so, DPM repositions computational approaches used to build knowledge in digital environments as tools to investigate the social, cultural, and communicational dynamics constructed in these spaces.
Using Algorithms to Study Algorithmic Cultures
HOW IT WORKS
Entity Extraction
The Digital Practices Matrix (DPM) identifies and normalizes entities such as hashtags, emojis, locations, posting times, profile categories, tagged accounts, or other desired parameters. This step organizes the data and prepares the variables for analysis, taking into account the usage grammar of the platforms.
Entity Relevance Analysis
The tool applies modified term weighting and information retrieval techniques to analyze the relevance of non-textual entities within groups. Adjustments in calculations and the incorporation of dynamic thresholds allow frequent entities to be considered without undervaluing them. This step categorizes usage patterns into three groups: exclusive, contrastive, and diluted practices.
Exclusive Practices: Patterns characteristic of specific groups, indicating differentiation strategies.
Contrastive Practices: Relevant practices within a group that are also present in others, highlighting convergences and distinctions in communication patterns across the corpus.
Diluted Practices: Communication patterns that do not characterize a specific group but may suggest more flexible and adaptable layers in the usage patterns of platform grammars.
Building the Matrix of Practices
The processed data is organized into a matrix that maps the distribution of practices across groups. This matrix also supports detailed reports and data visualizations, enabling the analysis of:
Practice Distribution: A broad overview of communication patterns within the corpus.
Group Characteristics: Insights into the combinations of digital practices that define the specific communication strategies of each group.
Entity Descriptions: A detailed breakdown of how each element contributes to these practices within groups, offering a granular perspective on what distinguishes seemingly similar practices or connects superficially contrasting ones.
Matrix Visualization
The matrix is displayed in an interactive format, allowing users to navigate the data and analyze how specific practices stand out across different groups. The interface simplifies the identification of patterns in concentration and distribution.
Export and Analysis
Results can be exported as detailed reports on practices by group and visualizations, providing empirical material for analyzing digital practices in various contexts. This supports comparative analyses, helps identify common or distinctive characteristics among groups, and anchors content analysis to the dynamics of visibility on digital platforms.
ADVANTAGES AND CUSTOMIZATION
Flexibility: DPM can be adapted to different corpora and categories, such as media types, profile biographies, or discussed topics.
Integration with Other Tools: Data extracted from network analysis, content analysis, topic modeling, or computational models can be aggregated and treated as categories for exploration.
Contextualization: Expands traditional analyses by integrating usage patterns, appropriations, and the dynamics of digital communication within groups as part of the interpretive process.
Flexibility and Context for Digital Analyses
POTENTIAL USES
Analyze the dynamics of digital practices on social networks like Instagram and TikTok.
Investigate social phenomena, such as disinformation campaigns or online protests.
Map communication and differentiation strategies among groups, such as digital influencers or activist communities.
Identify practices shared by groups, uncovering trends and convergences across various contexts.
Recognize adaptive patterns that reflect platform interface appropriations, usage cultures, and behaviors that can influence visibility dynamics and the reach of content and agendas online.
Influencers Polarization Usage Cultures
AVAILABILITY AND ACCESS
Currently, Digital Practice Matrix is under development and limited to ongoing research within the laboratory. It is not yet available for general use or broad distribution.
Collaborations may be considered for specific projects aligned with the lab's research agenda and team availability. Proposals for collaborations or partnerships will be reviewed based on their potential contribution to ongoing investigations.
If you are interested in exploring collaboration or investment opportunities, please contact the group leader for preliminary discussions: eliasbitencourt@gmail.com.