| Week |
Date, slides |
Summary, papers |
| 1 |
2010/01/05 |
- Guest lecture
on using clicks for community-driven search, by
Vijay Krishnan.
- Course summary and organization
- Finer-grain artifacts the Web: markups, token sequences,
user activity
- What forms of structure we want to identify:
records, entities, attributes, relations.
- What techniques to use: rule-based, statistical learning,
closed-domain, open-domain, etc.
- Why do this---use in search.
|
| 1 |
2010/01/08 |
- Regular patterns in (1d) tag sequences
- String edit distance
- Multiple string alignment
- Tree edit distance and alignment
|
| 2 |
2010/01/12 |
|
| 2 |
2010/01/14 |
(Guest lecture by Prof. Ganesh)
- Rule-based sequence labeling
|
| 3 |
2010/01/19 |
- Feature representation for tokens
- The B-I-O state trick and features
- MEMM
- CRF and training
- Quick intro to max-margin structured learning
|
| 3 |
2010/01/29 |
(Guest lecture by Prof. Sunita)
|
| 4 |
2010/02/02 |
(Guest lecture by Prof. Sunita)
|
| 4 |
2010/02/05 |
(Guest lecture by Prof. Sunita)
- Inference and Training with sequential CRFs
- Forward-backward recurrences
|
| 5 |
2010/02/09 |
(Guest lecture by Prof. Sunita)
|
| 5 |
2010/02/12 |
Shiv ratri holiday |
| 6 |
2010/02/20 |
Midterm exam,
SIC301, 14:00--17:00 |
| 7 |
2010/02/23 |
|
| 7 |
2010/02/26 |
|
| 8 |
2010/03/02 |
- Cucerzan's leave-one-out
- Milne and Witten's anchor method
|
| 8 |
2010/03/05 |
|
| 9 |
2010/03/09 |
Postponed |
| 9 |
2010/03/12 |
(Guest lecture by Prof. Sunita --
slides)
- Relationship extraction
- Clues: surface tokens, POS tags, parse tree, dependency graph
- Feature, rule, kernel -based methods
|
| 10 |
2010/03/16 |
Holiday |
| 10 |
2010/03/19 |
- More examples of path kernels
- Closed domain binary relationship bootstrapping:
DIPRE ,
Snowball
- Open-domain binary relation harvesting:
TextRunner
|
| 11 |
2010/03/23 |
- Coreference resolution:
in unstructured text, of rows in table
- Discriminative (active learning) approach
- Hierarchical hybrid network model
- Unified graphical model
|
| 11 |
2010/03/26 |
- Using entities and types in search
- Descendant of answer type in proximity to keyword matches:
IR4QA,
SSQ
- Extending to multiple answer types:
|
| 12 |
2010/03/30 |
|
| 12 |
2010/04/02 |
Holiday -- Good Friday |
| 13 |
2010/04/06 |
- Proximity-based (positional) language models in IR
- Probabilistic approaches to expert search
- Discriminative entity ranking
|
| 13 |
2010/04/09 |
(Guest lecture by Prof. Sudarshan)
- Proximity search in graphs
- DBXplorer ,
BANKS
(demo)
|
| 14 |
2010/04/13 |
- Summary of entity/expert search
- Convergence of DB+XML and IR+Web search
- WHIRL
- XIRQL and ELIXIR
- XQuery, TeXQuery
|
| 14 |
2010/04/16 |
|
| 15 |
2010/04/24 |
Final exam, SIC301, 14:30–17:30 |