Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Connectivity for Knowledge Building: A Framework of Socio-Semantic Network Motif Analysis

Connectivity for Knowledge Building: A Framework of Socio-Semantic Network Motif Analysis

Bodong Chen

June 06, 2022
Tweet

More Decks by Bodong Chen

Other Decks in Education

Transcript

  1. Connectivity for Knowledge Building: A Framework
    of Socio-Semantic Network Motif Analysis
    Bodong Chen
    ISLS Annual Meeting • June 6, 2022
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  2. Collaborative discourse
    Learners are engaged in discussing substantive content related to a domain
    ● social interaction, mediated communication, group awareness
    ● intersubjective meaning-making, transactivity, uptakes...
    Knowledge-building discourse engages students in productive discourse
    that leads towards continual improvement of ideas within a community
    (Scardamalia & Bereiter, 2014)
    2
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  3. Source
    See Chen & Hong, 2016, for a review
    Knowledge Forum
    3
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  4. Indicators of productive KB discourse
    Social network analysis; communication
    networks (e.g., Dawson, 2008; Haythornthwaite, 2015)
    Computational lexical, semantic, content
    analysis (Kovanović et al., 2017; Teplovs & Fujita, 2013)
    The social, cognitive, and integrated domains (Y Chen et al., 2019)
    Knowledge Connections Analyzer
    (Yang et al., 2021)
    Knowledge Building Discourse Explorer
    (KBDeX; Oshima et al., 2012)
    4
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  5. KBDeX, and its socio-semantic network analysis
    Learners
    Words
    Discourse
    Units
    5

    View Slide

  6. We need more integrative
    approaches to the analysis of
    collaborative discourse,
    as well as knowledge-building
    discourse in particular
    6
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  7. “The social and the cultural orders are
    dual – that is, they constitute each other……
    Socio-semantic network analysis brings
    together the study of relations among
    actors (social networks), relations among
    elements of actors’ cultural structures
    (their semantic networks), and relations
    among these two orders of networks.”
    — Basov et al. (2020)
    7
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  8. A Socio-Semantic Network Motifs Framework
    Socio-semantic networks (SSNs):
    ● the actors (learners) and semantic entities (words), along with
    their connections, are modeled as nodes and edges in a
    two-mode, dual-layer socio-semantic network
    Source 8
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  9. Socio-semantic networks
    Network motifs:
    ● Recurring, significant patterns of
    interconnections in the network
    (Milo et al., 2004)
    ● In network science, network
    motifs have been widely used to
    examine a variety of networks
    Socio-Semantic Network Motifs
    (Milo et al., 2002)
    9
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  10. Network motifs in
    socio-semantic
    networks
    Socio-Semantic Network Motifs
    10
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  11. Situate socio-semantic
    network motifs in
    collaborative discourse
    Socio-Semantic Network Motifs
    11
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  12. Situate socio-semantic
    network motifs in
    collaborative discourse
    Socio-Semantic Network Motifs
    12
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  13. Analyzing Socio-Semantic Network Motifs
    1) Construct the network
    ○ A range of analytical decisions
    (Chen & Poquet, 2022)
    ■ Learners: all learners in a class
    ■ Words: the top 100 high frequency
    words (after removing stopwords)
    which have appeared for at least 5
    times
    • Only writing behaviors
    • Threshold: edges with
    a weight >= 3
    Learner-Word
    • φ correlation
    coefficient >= .4
    Word-Word
    • Undirected edges, for
    simplicity
    Learner-Learner
    13
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  14. Analyzing Socio-Semantic Network Motifs
    1) Construct the network
    14
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  15. Analyzing Socio-Semantic Network Motifs
    1) Construct the network
    2) Count SSN motifs
    ○ count the occurrences of SSN motifs in the
    empirical network using the motifr R package
    (version 0.5.0)
    15
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  16. Analyzing Socio-Semantic Network Motifs
    1) Construct the network
    2) Count SSN motifs
    3) Compute significance of each motif
    ○ Generate 1,000 refined Erdos-Rényi random graphs
    ○ Compare the empirical network’s motif frequencies with the
    random graphs
    ○ A Z-score (-1, 1) is calculated for each SSN motif to show its
    over- or under-representation in the empirical network
    16
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  17. 17

    View Slide

  18. Example
    SSN motifs
    profile of a
    discourse
    segment:
    Analyzing Socio-Semantic Network Motifs
    18
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  19. ● A secondary dataset of knowledge-building discourse in 9th science
    ● Two contrasting classes:
    ○ Both classes showed progress across two phases around two topics
    ○ Class B engaged with more web resources, showed more intensive
    collaboration, and achieved greater progress on related topics
    Data
    Class Phase Posts Interactions Words/post
    Class A
    (n=22)
    1 68 33 77.2
    2 72 38 102.0
    Class B
    (n=26)
    1 97 40 65.4
    2 83 35 60.8
    19
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  20. Phase 1
    Phase 2
    Class A Class B 20

    View Slide

  21. 21

    View Slide

  22. Summary of findings
    Knowledge-building discourse in both classes showed considerable
    socio-semantic connectivity in both phases.
    SSN motifs showing exclusive access to words, e.g. (01,1a) and (11,1a),
    decreased, while the “hyper connectivity area” (upper-right corner) increased.
    Class A showed more room to improve socio-semantic connectivity in their
    discourse.
    22
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  23. Discussion
    We proposes a nascent socio-semantic network (SSN)
    motifs framework for the analysis of collaborative discourse.
    SSN motifs provide nuanced information about discourse.
    Ongoing work
    ● Adapt and evaluate the framework in different
    discourse contexts (e.g., Chen, Zhu, & Hong, 2022)
    ● Further combine SSN analysis with other methods
    ● Action-taking based on network motifs
    Link to slides: https://bit.ly/isls2022-motif

    View Slide

  24. Thank You! Acknowledgement
    This material is based upon work supported
    by the National Science Foundation under
    Grant No. 1657009.
    @bod0ng
    [email protected]
    Link to slides: https://bit.ly/isls2022-motif

    View Slide