search

kbastani / graphify

Graphify is a Neo4j unmanaged extension used for document and text classification using graph-based hierarchical pattern recognition.
http://graphify.github.io/graphify
Apache License 2.0
382 stars 89 forks source link

Training Step - Internal Server Error (500) #15

Closed cicero19 closed 9 years ago

cicero19 commented 10 years ago

Attempting to train the network using cURL requests. The first few worked, then I start getting status code: 500 and cannot add anymore and get the response below or a timeout.

HTTP/1.1 500 Server Error Content-Type: text/html; charset=ISO-8859-1 Cache-Control: must-revalidate,no-cache,no-store Content-Length: 0 Server: Jetty(9.0.z-SNAPSHOT)

Neo4j server is running fine. Can execute cyphers and view data.

Has this happened to anybody? Do you think it has anything to do with the data already added? Here is what it looks like untitled-1

kbastani commented 10 years ago

Can you take a look at the logs in Neo4j? Paste them here.

cicero19 commented 10 years ago

Hi Kenny,

Where exactly do I find this log file?

kbastani commented 10 years ago

Navigate to Neo4j's directory and go to /data/log. Copy the files here. Thanks!

cicero19 commented 10 years ago

I installed on linux and everything seems to be working properly. Must be something with that Windows installation. In fact, data/log does not exist in Windows installation and I am still unable to locate the log files.

Somewhat unrelated question: I have a graph with 2 classes (positive and negative) which I have trained with 3 examples of each. While I understand this is a very small amount of training data (I have 10,000 samples but just doing some preliminary testing now), when I try to classify an unlabeled body of text with http://localhost:7474/service/graphify/classify, I get this as the output:

{ "classes": [] }

ie. no classes given. Graph appears similar to the screenshot above. Is this simply a matter of too little training data or have I done something wrong in the training step?

Thanks for your help!

kbastani commented 9 years ago

Yes. Good question. When there is not enough information to confidently provide a set of labels for classification, no classes are returned.