Differences
This shows you the differences between two versions of the page.
| Next revision | Previous revision | ||
|
courses:rg:2012:atreport [2012/06/03 20:57] tamchyna vytvořeno |
courses:rg:2012:atreport [2012/06/03 21:01] (current) tamchyna added report, fixed format |
||
|---|---|---|---|
| Line 22: | Line 22: | ||
| - looking for the taxonomy that maximizes its likelihood given the evindence | - looking for the taxonomy that maximizes its likelihood given the evindence | ||
| - | 2.3 | + | ==== 2.3 ==== |
| - describes the main part of the algorithm | - describes the main part of the algorithm | ||
| - start with an initial taxonomy, add relations | - start with an initial taxonomy, add relations | ||
| Line 28: | Line 28: | ||
| - adding a relation involves adding relations implied by transitivity | - adding a relation involves adding relations implied by transitivity | ||
| - | 2.4 | + | ==== 2.4 ==== |
| - model adaptation for ambiguity | - model adaptation for ambiguity | ||
| - adding relations between word senses | - adding relations between word senses | ||
| - | Section 3 | + | ===== Section 3 ===== |
| - | 3.1 | + | ==== 3.1 ==== |
| - evidence is a vector | - evidence is a vector | ||
| - training data from WordNet, trained a classifier using logistic regression | - training data from WordNet, trained a classifier using logistic regression | ||
| - overfitting: | - overfitting: | ||
| - | 3.2 | + | ==== 3.2 ==== |
| - cousins | - cousins | ||
| - clustering -- similarity is cosine distance (within the cluster, 0 otherwise) | - clustering -- similarity is cosine distance (within the cluster, 0 otherwise) | ||
| - softmax regression: more than 2 classes, otherwise similar to logistic regression | - softmax regression: more than 2 classes, otherwise similar to logistic regression | ||
| - | 3.3 | + | ==== 3.3 ==== |
| - identify the set of word pairs that can be hyper/ | - identify the set of word pairs that can be hyper/ | ||
| - for each proposed hypernym... was explained | - for each proposed hypernym... was explained | ||
| - add the one with highest score according to the classifiers | - add the one with highest score according to the classifiers | ||
| - | 3.4 | + | ==== 3.4 ==== |
| - sense disambiguation | - sense disambiguation | ||
| - works by itself | - works by itself | ||
| - the " | - the " | ||
| - | Evaluation | + | ===== Evaluation |
| - manual evaluation -- uniformly generated samples from the first n links, human judge | - manual evaluation -- uniformly generated samples from the first n links, human judge | ||
| - annotators were to classify into 4 classes (4.1) | - annotators were to classify into 4 classes (4.1) | ||
| - all the various evaluation methods discussed | - all the various evaluation methods discussed | ||
| - | DISCUSSION: | + | ===== DISCUSSION |
| - what is MiniPar? very simple parser (that' | - what is MiniPar? very simple parser (that' | ||
| Line 65: | Line 65: | ||
| - features are both the dependency labels but they are also lexicalized: | - features are both the dependency labels but they are also lexicalized: | ||
| + | < | ||
| vehicle | vehicle | ||
| / | / | ||
| Line 72: | Line 73: | ||
| / | / | ||
| a | a | ||
| + | </ | ||
| results in features? maybe " | results in features? maybe " | ||