-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathreadme
42 lines (34 loc) · 2.31 KB
/
readme
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
These are the code and data for 'Joint Semantic Relevance Learning with Text Data and Graph Knowledge'
before running the code, please download data from "http://pan.baidu.com/s/1bn0LJun" and unzip it into ./data/
==Directory==
./data
corpus.4word2vec:training corpus for word vector.
test.animal-143: animal-143 test set/
test.sim-301: sim-301 test set. Note that each word in sim-353 may match several words in
WordNet due to ambiguity. Here in our experiments, for each
pair in sim-301, we compare the combination of all cases and
take similarity value of the most similar pair as its result.
train.wordnet-noun.pairs:graph training data of wordnet-noun.
train.wordnet-noun.wikipage.filter:joint text training data of wordnet-noun.
train.yago-animal.pairs:graph training data of yago-animal.
train.yago-animal.wikipage.filter:joint text training data of yago-animal.
train.wordnet-all.pairs:all the graph of Wordnet, which contains 15 types of relations.
train.wordnet-all.wikipage:joint text training data of all wordnet.
word2vec.100:the model we trained using word2vec with dimension 100.
word2vec.200:the model we trained using word2vec with dimension 200.
word2vec.50:the model we trained using word2vec with dimension 50.
word2vec.wordnet-noun.prt.100:the file of using word2vec to initialize entry vector of wordnet-noun, dimension 100.
word2vec.wordnet-noun.forpv.200:the file of using word2vec to initialize entry vector of wordnet-noun, dimension 200.
word2vec.wordnet-noun.forpv.50:the file of using word2vec to initialize entry vector of wordnet-noun, dimension 50.
word2vec.yago-animal.forpv.100:the file of using word2vec to initialize entry vector of yago-animal, dimension 100.
word2vec.yago-animal.forpv.200:the file of using word2vec to initialize entry vector of yago-animal, dimension 200.
word2vec.yago-animal.forpv.50:the file of using word2vec to initialize entry vector of yago-animal, dimension 50.
./src
main.cpp
==Compile==
"g++ src/main.cpp -o JTGR -lpthread"
==Implement Experiments==
"./JTGR" for instruction.
==NOTE==
The parameter setup of word2vec in our experiment is as followed,
"./word2vec -train corpus.4word2vec -output word2vec.dim100 -cbow 0 -size 100 -window 8 -negative 25 -hs 1 -sample 1e-4 -threads 10 -binary 0 -iter 10"