iunderstand / swe Goto Github PK

补充：使用word2vec训练词向量正常。

#hello，请问如何训练词向量？假设现在要训练SWE+Synon-Anton，我尝试如下步骤：

机器配置：Ubuntu 14.04 128G RAM

1. 提供wikipedia语料train.txt
2. 将 semantics/SWE.EN.KnowDB.WordNet-Book.Synon-Anton划分为sem.train.txt和sem.valid.txt
3. 运行
   ./SWE_Train -train train.txt -output vec.txt -size 200 -window 5 -sample 1e-4 -negative 5 -hs 0 -binary 0 -cbow 0 -iter 3 -sem-train sem.train.txt -sem-valid sem.valid.txt -sem-coeff 0.1 -sem-hinge 0.0 -sem-addtime 0 -weight-decay 0 -delta-left 1 -delta-right 1

出现问题：
读取语料后出现Segmentation fault！

#另外请问划分WordNet数据train和valid的比例？论文中并无提及

log如下：
Semantic Word Embedding (SWE) ToolkitTrain Setting embedding size: 200
Train Setting window size: 5
Train Setting sample value: 0.000100
Train Setting negative num: 5
Running Threads: 12
Iteration Times: 3
SemWE Qsem train file: ../semantics/SWE.EN.KnowDB.WordNet-Book.Synon-Anton.train
SemWE Qsem valid file: ../semantics/SWE.EN.KnowDB.WordNet-Book.Synon-Anton.valid
SemWE Add Time(/%): 0.000000
SemWE Weight Decay: 0.000000
SemWE Inter Coeff: 0.100000
SemWE Norm Hinge Margin: 0.000000
SemWE Inequation Delta Left: 1
SemWE Inequation Delta Right: 1
Training Starting @time: Fri Feb 24 19:21:08 2017

Starting training using file wikicorpus.1b
Vocab size: 218317
Words in train file: 123353508
Load Training Word Knowledge from file ../semantics/SWE.EN.KnowDB.WordNet-Book.Synon-Anton.train
--- InEquation Nums: 424732
--- Finish reading the Knowledge Database
Load CV Test Word Knowledge from file ../semantics/SWE.EN.KnowDB.WordNet-Book.Synon-Anton.valid
--- CV set InEquation Nums: 1000
./run.sh: line 5: 25479 Segmentation fault (core dumped) ./SWE_Train -train ${TRAIN_FILE} -output vec.bin -size 200 -window 5 -sample 1e-4 -negative 5 -hs 0 -binary 1 -cbow 0 -iter 3 -sem-train ${SEW_FILE} -sem-valid ${SEW_CV_FILE} -sem-coeff 0.1 -sem-hinge 0.0 -sem-addtime 0 -weight-decay 0 -delta-left 1 -delta-right 1

The idea is excellent!
1. Could you provide some clues how these set synon-anton,
Hyper-Hypon (e.g "../SWE.EN.KnowDB.WordNet-Book.Synon-Anton" ) extracted the ontology such as WordNet?
According to the Semantic Category Rule in your document, "A semantic category may be defined as a synset in WordNet, a hypernym in a semantic hierarchy, or an entity category in knowledge graphs." Is the synset used come from WordNet or get extracted by you?
2. Contrary to the example "similarity(Mallet, Hammer) > similarity(Mallet, Tool)" Does all words in knowledge graph need to be considered ?? e.g similarity(Mallet, Hammer) > similarity(Mallet, WordFaraway)
Thanks for your sharing!

When I run the SWE_Train using the following command:

./SWE/bin/SWE_Train -debug 2 -size 100 -train ./corpora/corpus.txt -read-vocab ./corpora/vocabulary.txt -cbow 0 -hs 0 -alpha 0.025 -window 5 -sample 0.0001 -negative 5 -threads 1 -output ./word_embed.txt -sem-coeff 0.005 -sem-addtime 0 -sem-hinge 0.0 -weight-decay 0.0 -sem-train ./semantics/knowledge_constraints.train -sem-valid ./semantics/knowledge_constraints.valid -iter 2

The ./semantics/knowledge_constraints.train and ./semantics/knowledge_constraints.valid are the same file as SemWE.EN.KnowDB.COM1.inTEXT8.train and SemWE.EN.KnowDB.COM1.inTEXT8.valid in semantics/TEXT8 directory.

The output I got is:

Semantic Word Embedding (SWE) ToolkitTrain Setting embedding size: 100
Train Setting window size: 5
Train Setting sample value: 0.000100
Train Setting negative num: 5
Running Threads: 1
Iteration Times: 2
SemWE Qsem train file: ./semantics/knowledge_constraints.train
SemWE Qsem valid file: ./semantics/knowledge_constraints.valid
SemWE Add Time(/%): 0.000000
SemWE Weight Decay: 0.000000
SemWE Inter Coeff: 0.005000
SemWE Norm Hinge Margin: 0.000000
SemWE Inequation Delta Left: 1
SemWE Inequation Delta Right: 1

Training Starting @time: Sat Nov 11 12:04:57 2017

Starting training using file ./corpora/corpus.txt
Vocab size: 47091
Words in train file: 9614559

Load Training Word Knowledge from file ./semantics/knowledge_constraints.train
--- InEquation Nums: 324817
--- Finish reading the Knowledge Database
Load CV Test Word Knowledge from file ./semantics/knowledge_constraints.valid
--- CV set InEquation Nums: 2999
--- Finish reading the CV Knowledge Database
--- Alpha: 0.025000 Progress: 0.00% WordCount: 0 Train_Qsem: inf Train_SatisfyRate: 0.0000 Valid_Qsem: inf Valid_SatisfyRate: 0.0000
Segmentation fault

Did I use the SWE_Train incorrectly? Btw, I suggest you provide a documentation for explaining how to use the SWE_Train file.

Thanks for your help

你好，我运行task1的perl SWE_Train_TEXT8.pl时候出现了这个错误。不知道是为什么。希望可以得到您的帮助