def test(n_docs = 100):
db = MongoClient().patents
w2v,kmeans = model_loader(300,200)
def part_func(doc):
return {'$set': {'wordvec_clusters': cluster_distances(db, doc['_id'], w2v,kmeans)}}
for doc in db.traits.find({'doc_vec': {'$exists': True, '$nin': [[0 for _ in range(300)]]}, 'top_tf-idf': {'$nin': [[]]}}).limit(n_docs):
pprint(part_func(doc))
def test():
db = MongoClient().patents
w2v,kmeans = model_loader(300,200)
dists = {name: cluster_distances(db, pno, w2v,kmeans)
for (name,pno) in _friendly_patents
}
return w2v, kmeans, dists, parse_clusters(kmeans, w2v)
def main():
db = MongoClient().patents
w2v, kmeans = model_loader(300, 200)
def part_func(doc):
return {
'$set': {
'wordvec_clusters': cluster_distances(db, doc['_id'], w2v,
kmeans)
}
}
parallelMap(part_func,
in_collection=db.traits,
out_collection=db.traits,
findArgs={
'spec': {
'doc_vec': {
'$exists': True,
'$nin': [[0 for _ in range(300)]]
},
'top_tf-idf': {
'$nin': [[]]
}
},
'fields': {
'_id': 1
}
},
updateFreq=500,
bSize=1000)
def test():
db = MongoClient().patents
w2v, kmeans = model_loader(300, 200)
dists = {
name: cluster_distances(db, pno, w2v, kmeans)
for (name, pno) in _friendly_patents
}
return w2v, kmeans, dists, parse_clusters(kmeans, w2v)
def test(n_docs=100):
db = MongoClient().patents
w2v, kmeans = model_loader(300, 200)
def part_func(doc):
try:
return {"$set": {"wordvec_clusters": cluster_distances(db, doc["_id"], w2v, kmeans)}}
except:
return {"$set": {"wordvec_clusters": []}}
for doc in db.traits.find().limit(n_docs):
pprint(part_func(doc))
def main():
db = MongoClient().patents
w2v,kmeans = model_loader(300,200)
def part_func(doc):
return {'$set': {'wordvec_clusters': cluster_distances(db, doc['_id'], w2v,kmeans)}}
parallelMap(
part_func,
in_collection = db.traits,
out_collection = db.traits,
findArgs = {
'spec': {'doc_vec': {'$exists': True, '$nin': [[0 for _ in range(300)]]}, 'top_tf-idf': {'$nin': [[]]}},
'fields': {'_id': 1}
},
updateFreq=500,
bSize=1000
)
def main():
db = MongoClient().patents
w2v, kmeans = model_loader(300, 200)
def part_func(doc):
try:
return {"$set": {"wordvec_clusters": cluster_distances(db, doc["_id"], w2v, kmeans)}}
except:
return {"$set": {"wordvec_clusters": []}}
parallelMap(
part_func,
in_collection=db.traits,
out_collection=db.traits,
findArgs={"spec": {}, "fields": {}},
updateFreq=500,
bSize=1000,
)
def test(n_docs=100):
db = MongoClient().patents
w2v, kmeans = model_loader(300, 200)
def part_func(doc):
return {
'$set': {
'wordvec_clusters': cluster_distances(db, doc['_id'], w2v,
kmeans)
}
}
for doc in db.traits.find({
'doc_vec': {
'$exists': True,
'$nin': [[0 for _ in range(300)]]
},
'top_tf-idf': {
'$nin': [[]]
}
}).limit(n_docs):
pprint(part_func(doc))
dists = {name: cluster_distances(db, pno, w2v_model, cluster_model)
for name,pno in pnos}
cluster_parse_fn = '/'.join([outdir, 'parsed_clusters.p'])
dist_fn = '/'.join([outdir, 'patent_dists.p'])
tsne_fn = '/'.join([outdir, 'embedding_fig_tsne.png'])
save_dict(cluster_parse_fn, parsed_clusters)
save_dict(dist_fn, dists)
embedding_fig(w2v_model, cluster_model, savefn = tsne_fn, n=150)
def test():
db = MongoClient().patents
w2v,kmeans = model_loader(300,200)
dists = {name: cluster_distances(db, pno, w2v,kmeans)
for (name,pno) in _friendly_patents
}
return w2v, kmeans, dists, parse_clusters(kmeans, w2v)
if __name__ == '__main__':
if len(sys.argv) != 4:
exit("Usage: python {} <wordvec_size> <n_clusters> <out directory>".format(sys.argv[0]))
db = MongoClient().patents
w2v_dim = int(sys.argv[1])
n_clusters = int(sys.argv[2])
outdir = sys.argv[3]
try:
w2v,kmeans = model_loader(w2v_dim, n_clusters)
except:
exit("Model with vector dimension {} and {} clusters not found.".format(w2v_dim, n_clusters))
model_report(db, _friendly_patents, w2v, kmeans, outdir)
save_dict(cluster_parse_fn, parsed_clusters)
save_dict(dist_fn, dists)
embedding_fig(w2v_model, cluster_model, savefn=tsne_fn, n=150)
def test():
db = MongoClient().patents
w2v, kmeans = model_loader(300, 200)
dists = {
name: cluster_distances(db, pno, w2v, kmeans)
for (name, pno) in _friendly_patents
}
return w2v, kmeans, dists, parse_clusters(kmeans, w2v)
if __name__ == '__main__':
if len(sys.argv) != 4:
exit("Usage: python {} <wordvec_size> <n_clusters> <out directory>".
format(sys.argv[0]))
db = MongoClient().patents
w2v_dim = int(sys.argv[1])
n_clusters = int(sys.argv[2])
outdir = sys.argv[3]
try:
w2v, kmeans = model_loader(w2v_dim, n_clusters)
except:
exit(
"Model with vector dimension {} and {} clusters not found.".format(
w2v_dim, n_clusters))
model_report(db, _friendly_patents, w2v, kmeans, outdir)
