emotions, experiences and social media - world wide …€¢ intelligent information access •...
TRANSCRIPT
Emotions, Experiences and Social Media
Maarten de Rijke
University of Amsterdam
1
Another perspective
• Standards, infrastructure as seen by an academic research group
• Intelligent information access• Content-based matching
• Additional features (recency, authoritativeness, novelty, opinionatedness, …)
• Combine content-based and additional features
• Presentation
2
2
Research strategy
3
Theory
Experiment
Application
3
Online lives
4
Why?
4
What?
5
5
What?
5
A web of applications
Social media analysis
Learning from implicit feedback
Reputation management: identifying and tracking stakeholders
Finding experiences to inform creation of new products
Real-time impact prediction
Multilingual log file analysis
Interest machine
Email search
Computational humanities and social sciences
Linking archives
Providing access to all parliamentary data in Europe
News archives meet video archives meet …
Open data initiatives
Religious studies
Medical anthropology
Communication science Governmental
Chronobiology
E-discovery
Detecting radicalization
Entity mining
Enrichment through linking
Aggregated entity search
5
Political Mashup
• Aggregating parliamentary data
• Debates, debate structure
• “Semantification”
• Linking to video broadcasts, twitter, blogs, party programs
• Tracking topic ownership from parliament to social media and back
6
CoSyne• Translate between wiki
pages
• Identify changes in one page
• Find gaps in other, target pages
• Translate material to be inserted in gaps
• Insert translated material in gaps
7
• Mood annotated blogs
• Real-time mood tracking and prediction
• MoodViews (2005-2009)• Moodgrapher: follow
• Moodteller: predict
• Moodsignals: explain
• Moodspotter: discover associations
• Analyzing ‘old’ data: chronobiology
The mood of the web
8
8
• Mood annotated blogs
• Real-time mood tracking and prediction
• MoodViews (2005-2009)• Moodgrapher: follow
• Moodteller: predict
• Moodsignals: explain
• Moodspotter: discover associations
• Analyzing ‘old’ data: chronobiology
The mood of the web
http://www.moodviews.com8
8
• Mood annotated blogs
• Real-time mood tracking and prediction
• MoodViews (2005-2009)• Moodgrapher: follow
• Moodteller: predict
• Moodsignals: explain
• Moodspotter: discover associations
• Analyzing ‘old’ data: chronobiology
The mood of the web
http://www.moodviews.com8
8
• Mood annotated blogs
• Real-time mood tracking and prediction
• MoodViews (2005-2009)• Moodgrapher: follow
• Moodteller: predict
• Moodsignals: explain
• Moodspotter: discover associations
• Analyzing ‘old’ data: chronobiology
The mood of the web
http://www.moodviews.com8
8
• Mood annotated blogs
• Real-time mood tracking and prediction
• MoodViews (2005-2009)• Moodgrapher: follow
• Moodteller: predict
• Moodsignals: explain
• Moodspotter: discover associations
• Analyzing ‘old’ data: chronobiology
The mood of the web
http://www.moodviews.com8
8
• Mood annotated blogs
• Real-time mood tracking and prediction
• MoodViews (2005-2009)• Moodgrapher: follow
• Moodteller: predict
• Moodsignals: explain
• Moodspotter: discover associations
• Analyzing ‘old’ data: chronobiology
The mood of the web
http://www.moodviews.com8
06/03/05 08/01/05 10/01/05 12/01/05 02/01/06 04/01/060.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Original data versus Trend
ratio
of
blo
g p
osts
la
be
led
with
ST
RE
SS
ED
8
Ingredients
• Search engine technologies
• Content extraction
• Language technologies
• Semistructured data technologies
• Scaleable distributed processing
9
9
• We are scientists, developers, users at the same time and we have external partners
• Agile vs standards?
• Let a 1000 flowers bloom?
Development strategy
10
10
vs.
11
Fietstas
12
FietstasProgrammer API
Python
SscrapeFietstas web scraperfor RSS feeds or sites
WWW
Fietstas
Fietstas worker
Stemming, NEN, NER, term cloud aggregation, ...
Fietstas Web APIXMLRPC or REST
Fietstas InspectorInspect document
annotations
direct document upload
file by file or in batches
Fietstas worker
Stemming, NEN, NER, term cloud aggregation, ...
Fietstas worker
Stemming, NEN, NER, term cloud aggregation, ...
documents
Text analysis service (NL, EN)
12
• A look from the lab
• Social media as a “societal thermometer”
• Many opportunities for public-private collaborations
• Infrastructure for supporting these collaborations
13
• Based on joint work with
• Krisztian Balog, Wouter Bolsterlee, Breyten Ernsting, Valentin Jijkoun, Fons Laan, Maarten Marx, Gilad Mishne, Christof Monz, Daan Odijk, Ork de Rooij, Manos Tsagkias, Andrei Vishneuski
14