Transcript The presentation template

School of Computing
something
FACULTY OF ENGINEERING
OTHER
A research-led coursework assignment for the
exceptional student and the student journey:
Co-author an interdisciplinary research paper!
Based on British Computer Society – Machine Intelligence contest entry:
Combinatory Hybrid Elementary Analysis of Text:
hybrid human-machine intelligence to generate
research journal papers by Eric Atwell et al.
Research Institute for Artificial Intelligence and Biological Systems,
School of Computing, Faculty of Engineering, Leeds University
Outline
• Introduction
• Experiments to combine research and teaching in AI / Knowledge
Management and Corpus Linguistics, using an AI-inspired intelligent
agent architecture, but casting students as the intelligent agents. Each
student applies KM/Data-mining to a corpus, then we combine results
• Methods
• Students given detailed coursework spec, write up as a research paper
• Results
• Draft research papers by Junaid Arshad, Chien-Ming Lai, Lan Nim,
Noushin Rezapour Asheghi, Josiah Wang, Justin Washtell ( + more?)
• Conclusions
• Spamming journals? More research questions for next year’s classes
INTRODUCTION:
controversial assumptions?
Q: What’s the greatest commercial success on the Internet?
INTRODUCTION:
controversial assumptions?
Q: What’s the greatest commercial success on the Internet?
A: not PORN ... but ADVERTISING!
(not EskimoTube, but YouTube and other Google adverts)
Google is a web-search service?
- NO: Google is an advert-generating service!
- even EskimoTube makes money through adverts
INTRODUCTION:
controversial assumptions?
Q: What’s the greatest commercial success on the Internet?
A: not PORN ... but ADVERTISING!
(not EskimoTube, but YouTube and other Google adverts)
SPAM is a particularly successful innovation: generating large
numbers of personalised adverts for potential customers
Spam WORKS: even if only a low “success rate” ...
INTRODUCTION:
controversial assumptions?
Q: What is the aim of academic research?
INTRODUCTION:
controversial assumptions?
Q: What is the aim of academic research?
A: The aim of academic research is to generate journal papers
(for RAE, for publicity, for promotion, ?)
RAE: Researchers must produce 4 research papers in 6 years
Students + Machine Intelligence: 40 draft papers in 6 weeks
-a BIG advance in Machine Intelligence
- AND research-led student learning
We integrate AI and Human Intelligence at 3 levels:
using AI architecture; AI tools; AI students
Background assumptions
The aim of research is to generate conference/journal papers
(for RAE, for publicity, for promotion, ?)
Computing students should learn to apply ICT to practical,
“real” / “useful” tasks
Research-led teaching and learning is a Leeds Univ strength
SO … students could learn by applying ICT to research
questions, and writing research papers on results?
BUT … research is “hard” – surely a student can’t come up
with ideas and results for a publishable research paper?!
Maybe one student can’t … but…
Intelligent Agent Architecture
Wikipedia: “In computer science, an intelligent agent (IA) is a
software agent that exhibits some form of artificial
intelligence that assists the user and will act on their behalf,
in performing repetitive computer-related tasks. While the
working of software agents used for operator assistance or
data mining (sometimes referred to as bots) is often based
on fixed pre-programmed rules, "intelligent" here implies the
ability to adapt and learn … a multi-agent system (MAS) is
a system composed of several agents, collectively capable of
reaching goals that are difficult to achieve by an individual
agent or monolithic system … A multiple agent system
(MAS) is a distributed parallel computer system built of many
very simple components, each using a simple algorithm, and
each communicating with other components. A paradigm of
an ant colony or bee swarm is used many times.”
INTRODUCTION:
Students as intelligent agents
Monkeys at typewriters will eventually type Shakespeare plays
Bio-Inspired Computing researchers aim to develop software
which behaves like ants, bees, etc to achieve complex results
Why not use students as “super-intelligent agents”??
Prof David Cliff: this is “cheating” – his goal is software agents
BUT my goal is to generate research journal papers,
I am not aiming to build bio-inspired computing software!
METHODS
Structure: Introduction, Methods, Results, Conclusions.
I can’t tell you the details in 10 minutes…
Instead… a bird’s eye view: how student + MI can help me
generate a draft journal paper
I am the QB “queen bee”: I guide the hive (students+MI)
We have 10 minutes, not 6 weeks, so key steps only…
NB: COMPUTING DEMOS ALWAYS GO WRONG
METHOD:
How to create a journal paper
QB) Design the overall HI-MI hybrid: coursework
specification:
http://www.comp.leeds.ac.uk/eric/cr32/cr32cw1.doc
http://www.comp.leeds.ac.uk/eric/wwe.shtml
QB) Select a domain + research question where MI
is novel
Humanities: Language and Cultural studies for a
region; specifically: Which English dominates
WWW in this region, British or American?
METHOD:
How to create a journal paper
Which English dominates WWW in your region,
British or American?
1) Use AI search tool to choose a region and journal for this
question; and find related research to cite, in the
Introduction of your paper.
2) Choose 3+ countries in this region, use AI search tool to
harvest a Web-Corpus for each country
QB) harvest 10 UK and 10 US Web-corpus data-samples
How to create a journal paper
(continued…)
QB) Use AI tool to find significant differences:
ML features characteristic of UK v. US English
3) Choose features, encode in uk-us ARFF file
4) Chosen region: encode features in test ARFF file
5) Use AI ML toolkit (WEKA) to build visualisation
and ML evidence of uk-us decision;
copy-and-paste into journal paper: novel evidence
(novel for this readership!)
How to create a journal paper
(continued…)
6) Decision-tree predictions for region samples:
UK or US? (Test options: Supplied test set);
copy into journal paper
7) Finish paper: Introduction, Methods, Results …
(ML evidence: novel to this research journal
readership), Conclusions
8) Submit paper via AI Knowledge Management tool
QB) assess course-works, aka review/improve,
submit to real conference/journal
RESULTS
Student: outline paper for co-author to expand; great for CV!
QB) 40 research papers to submit to conferences and journals
… I’ll let you know how many were accepted for
Language Resources and Evaluation Conference 2010, Malta
CONCLUSIONS
For BCS Machine Intelligence contest:
A hybrid of human and machine intelligence -
AI architecture applied to students + smart choice of journals
and instructions + use of AI tools by AI students
… can produce 40 draft research papers in 6 weeks
– much more than HI or MI alone.
-a BIG advance in Machine Intelligence
- AND Human Intelligence – novel student learning experience
CONCLUSIONS
What else?...
Students to draft research-led teaching resources for
Wikipedia University: Wikiversity
Same procedure, but instead of a “paper”, write a Wiki entry
Not just AI students: any computer-literate student (or
lecturer!) can use Wikipedia, and add to Wiki
A.I. Lecturers could collaborate with Language/Humanities
Lecturers to produce sharable “blended learning” resources
Eg at UPenn: Natural Language Tool Kit NLTK
At Leeds: Quranic Arabic Corpus
LTC7: http://www.comp.leeds.ac.uk/eric/atwell10ltc.ppt
Interested? Google me: “Eric Atwell”