Scribd Logo
Upload

Welcome to Scribd!

  • 21 views

    WP3 Challenges & Hybrid Models: Dan Brickley, VUA Pro-Netics & BBC

    Uploaded by

    DanBrickley
    This document discusses challenges and hybrid models for recommending TV content using social web data. It addresses issues of data sparsity, privacy, and heterogeneity across different platforms. It describes building different types of recommenders, including a pattern recommender that finds relationships between users and content, and a collaborative filtering recommender. Plans include further integrating these approaches and analyzing emerging standards and big social data.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd

    WP3 Challenges & Hybrid Models: Dan Brickley, VUA Pro-Netics & BBC

    Uploaded by

    DanBrickley
    21 views30 pages
    This document discusses challenges and hybrid models for recommending TV content using social web data. It addresses issues of data sparsity, privacy, and heterogeneity across different platforms. It describes building different types of recommenders, including a pattern recommender that finds relationships between users and content, and a collaborative filtering recommender. Plans include further integrating these approaches and analyzing emerging standards and big social data.

    Original Title

    Danbri 2nd Review WP3 Challenges

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PPT, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses challenges and hybrid models for recommending TV content using social web data. It addresses issues of data sparsity, privacy, and heterogeneity across different platforms. It describes building different types of recommenders, including a pattern recommender that finds relationships between users and content, and a collaborative filtering recommender. Plans include further integrating these approaches and analyzing emerging standards and big social data.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    21 views30 pages

    WP3 Challenges & Hybrid Models: Dan Brickley, VUA Pro-Netics & BBC

    Uploaded by

    DanBrickley
    This document discusses challenges and hybrid models for recommending TV content using social web data. It addresses issues of data sparsity, privacy, and heterogeneity across different platforms. It describes building different types of recommenders, including a pattern recommender that finds relationships between users and content, and a collaborative filtering recommender. Plans include further integrating these approaches and analyzing emerging standards and big social data.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    of 30

    WP3

    Challenges & Hybrid models


    Dan Brickley, VUA
    Pro-netics & BBC
    Overview
    • Challenges for TV in Social Web
    – theory and practice of our hybrid approach
    • 3 Interconnected problems:
    – Privacy, Sparsity and Heterogeneity
    • What we built (and why)
    – different kinds of recommender
    – ways of integrating them
    • Plans and options for final developments

    2
    Theory and Practice

    9 8 5 2 0 9

    0 0 8 8 8 6

    3 2 7 9 9 8

    3
    More likely ...

    9 0 0 0 0 9

    0 0 8 0 8 0

    0 0 7 0 9 8

    4
    5
    6
    7
    8
    TV preference data is very sparse!
    • Even for a single service (eg. Netflix), data is
    ‘overwhelmingly sparse’
    • For NoTube’s open systems, challenges
    multiply:
    – often no global view, only per-user data
    – many ways of identifying the same content item
    – many ways of identifying the same user
    – never mind other entities (actors, directors, ...)

    9
    Challenges: Sparsity, Fragmentation

    • Content identifiers (WP1)


    – Wikipedia/DBpedia URLs? Freebase?
    – RottenTomatoes.com, IMDB.com, broadcaster IDs
    • Social Web interoperability
    – Bob’s on Facebook, Charlie’s on Twitter
    – negotiating access to non-public data (OAuth)
    – reconciling metadata models, rating models
    10
    Fragmentation by site

    11
    A hybrid approach to sparsity

    • Find patterns and paths in factual data


    • Collaborative filtering - from bulk rating data
    • Experiments with ‘big data’ (e.g. Twitter
    crawl)
    • Models for combining recommenders
    • Strategies for inferring ‘sameAs’ links
    • ...or grouping items together (by series,
    brand)
    12
    Challenge: Privacy
    • TV preferences are very personal data
    • Relevant standards (OAuth) are new
    – deployed widely in Social Web during NoTube
    – slower adoption in TV and broadcast world
    • We can use OAuth to request permission to
    read a user’s closed data (eg. Facebook ‘Like’s)
    • limits ability to find general trends across an
    entire audience (except public data - twitter?)

    13
    Diversity and Fragmentation
    • Diversity of the Web
    – reading lists: bookcrossing, librarything, amazon
    – music on last.fm, spotify, ...
    – news sites, social networks, blogs ...
    • How to integrate while respecting privacy?
    • Good news: OAuth deployment growing &
    social sites expose their recommendations
    • Bad news: user-by-user data makes large-scale
    analysis of trends harder
    14
    OAuth? RDFa?
    • OAuth lets sites negotiate access with users
    • e.g., Facebook knows lots of movies I “like”.
    • NoTube can use OAuth to ask me to share that
    data with TV services
    • RDFa data from movie pages (IMDB, Rotten
    tomatoes) is consumed at Facebook
    • This makes certain pages attractive as content
    identifiers, a ‘taste graph’ alongside ‘social
    graph’
    15
    16
    17
    RDFa in IMDB and
    RottenTomatoes HTML

    Aggregated by Facebook (and then, by us...) 18


    What we built
    • Main WP3 work: beancounter and pattern
    recommender
    • Aggregate, normalize and merge social Web
    activity streams, then match against enriched
    TV metadata to produce recommendations
    • We also have a Mahout-based collaborative
    filtering recommender, with ‘item to item’
    recommendations based on bulk ratings data

    19
    LOD challenges
    • Linked Open Data for TV is new
    – datasets evolving, changing
    – quality varies
    – modelling styles vary
    – ‘lumpy’, uneven coverage
    • ‘Pattern recommender’ finds paths
    – from items in user profile to new content
    – handles variation between Linked Data sources

    20
    Content Pattern-based
    Recommendations
    • Paths in Linked Open Data
    • Diversity & Serendipity measures

    21
    Participation Pattern
    • Person X played role Y
    in TV program Z

    • 194,649 lmdb:actor triples


    • 53,180 lmdb:director triples
    • 28,549 lmdb:writer triples
    • 1,262
    lmdb:film_story_contributor
    triples

    22
    Influence Pattern
    • Person X influenced
    by person Y (direct)
    • Person X and Y
    influenced by person
    Z (in-direct)

    • 6,562 dbpedia:influenced
    triples
    • 11,776 dbpedia:influencedBy
    triples

    23
    Analysis of Patterns in Dataset
    # items # items
    recommendations 1266
    recommendations 222
    - Individual brands 411
    - Individual brands 100
    paths 17,001
    paths
    - with linkedmdb:actor 15,257
    - with linkedmdb:director 1155 - influencedBy (all) 1202

    - with linkedmdb:writer 569


    - influencedBy (unique) 521
    - with linkedmdb:film_story_contributor 20

    Dataset (BBC EPG metadata):


    – 12,777 (7,756 title enrichment) programmes
    – 1260 (401 enriched) brands (unique titles)
    – 35,227 (19,394 enriched) person names in metadata
    – 9,315 (4,590 enriched) unique person names in metadata
    24
    Collaborative filtering

    (item similarity measures from bulk ratings data)


    25
    26
    27
    Hybrid models:
    factual paths and statistical similarity

    (and not to mention ‘@wossy’ is on Twitter with 1 million followers...)


    28
    Status
    • We can show a standards-based system that
    – integrates TV preference data from diverse Web
    – matches this with enriched TV metadata
    – finds graph patterns linking users to content
    – integrates with ‘classic’ recommender approaches
    – builds on opensource (Cliopatria, Mahout)
    – supports real-time multi-screen exploration

    29
    Plans and challenges
    • Richer integration between components
    – currently this occurs in the application; can we
    exploit LOD patterns prior to Mahout analysis?
    • Polish & packaging; more patterns and rules
    • Track and influence evolving standards (W3C)
    • Work-in-progress with ‘big data’ analysis -
    ‘what kinds of TV links are shared by the kind
    of people who follow @stephenfry on
    Twitter?’
    30

    You might also like