-
Notifications
You must be signed in to change notification settings - Fork 0
/
semweblib.py
581 lines (509 loc) · 21.4 KB
/
semweblib.py
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
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
#semantic node, ie word plus semantic meaning representation
import hashlib
import _pickle as pickle
import inspect#needed?
import dateLib
import json
import nltk
from nltk.corpus import stopwords
nltk.download('wordnet')
#-------------- aidan's notes --------------------
#layer of action TRACKS in web *****!!!
#sem_edges have hash of two connecting words creating unique key for pairs of words
#perhaps a higher level edge for a->root->b
#don't redraw conflicting facts or questions
#-------------------------------------------------
def freeze_web(web_cl):
fh = open("memory/serialized-instances/chillyWeb.obj", 'wb')
pickle.dump(web_cl, fh)
def torch_web():
web_cl = sw()
fh = open("memory/serialized-instances/chillyWeb.obj", 'wb')
pickle.dump(web_cl, fh)
def thaw_web():
torch_web()
fh = open("memory/serialized-instances/chillyWeb.obj", 'rb')
wewb = pickle.load(fh)
return(wewb)
class sem_node:
#'semantic hash' function
#uniquely represents BOTH text and semantic form/meaning
#superior to text as words which take different forms have different meanings
def semHasher(self):
self.semHash = hashlib.md5((self.text+self.form).encode()).hexdigest()
def __init__(self, stringinit, POS, opos):
self.inbound_edges = []
self.outbound_edges = []
self.node_x = None
self.node_y = None
#textual representation
self.text = stringinit
#semantic POS
self.form = POS
self.alt_form = opos
self.entity_tag = 'none'
self.qual = "node"
self.bind = None
#semantic hash initialization
self.semHash = None
self.semHasher()
#profile link
#allows for direct resolution of subjects to other pieces of information
self.profileLink = None
self.individual_traces = []
class live_node:
def semHasher(self):
self.semHash = hashlib.md5((self.type).encode()).hexdigest()
def __init__(self, stringinit, type):
self.inbound_edges = []
self.outbound_edges = []
self.node_x = None
self.node_y = None
#textual representation
self.text = stringinit
self.qual = "live_node"
self.type = type
self.entity_tag = self.type
#semantic hash initialization
self.semHash = None
self.semHasher()
#semantic POS (not relevant here)
self.form = "live"
self.alt_form = "none"
self.profileLink = None
#individual_traces
self.individual_traces = []
class live_vector:
def semHasher(self):
self.semHash = hashlib.md5((self.type).encode()).hexdigest()
def __init__(self, stringinit, type):
self.inbound_edges = []
self.outbound_edges = []
self.node_x = None
self.node_y = None
#textual representation
self.text = stringinit
self.qual = "live_node"
self.type = type
self.entity_tag = self.type
#semantic hash initialization
self.semHash = None
self.semHasher()
#semantic POS (not relevant here)
self.form = "live"
self.alt_form = "none"
self.profileLink = None
self.conversation_tag = None
#individual_traces
self.individual_traces = []
class entity_node:
def semHasher(self):
self.semHash = hashlib.md5((self.type).encode()).hexdigest()
def __init__(self, stringinit, type):
self.inbound_edges = []
self.outbound_edges = []
self.node_x = None
self.node_y = None
#textual representation
self.text = stringinit
self.qual = "entity_node"
self.type = type
self.entity_tag = self.type
#semantic hash initialization
self.semHash = None
self.semHasher()
#semantic POS (not relevant here)
self.form = "entity"
self.alt_form = "none"
self.profileLink = None
#individual_traces
self.individual_traces = []
#carries meta-data for vector,
class sem_vector:
def __init__(self):
self.speaker = None
self.frame = None
#time metadata
self.t = 0.0
#sentence type
self.type = None
#actual vector
self.track = []
self.text = None
self.chunk_indices = None
self.entity_indices = None
self.subj = None
self.obj = None
self.relation = None
self.tense = None
def resolve_chunk_indices(self):
#print(self.frame['chunks'])
ci = []
chunks = self.frame['chunks']
for x in range(0, len(chunks)):
cheld = chunks[x].split(' ')
for y in range(0, len(cheld)):
chunks.append(cheld[y])
if(self.frame['chunks']==None):
self.chunk_indices = ci
elif(self.chunk_indices != None):
if(len(self.chunk_indices)==0 or len(self.resolve_chunk_indices) >= 1):
return(self.chunk_indices)
else:
for x in range(0, len(self.track)):
if(self.track[x].qual == 'node'):
if(self.track[x].text in chunks):
ci.append(x)
self.chunk_indices = ci
print(chunks)
return(self.chunk_indices)
def entify(self):
#type_lookup = {"CARDINAL":0,"DATE":2,"EVENT":4,"FAC":6, "GPE":8, "LANGUAGE":10, "LAW":12, "LOC":14, "MONEY":16, "NORP":18, "ORDINAL":20, "ORG":22, "PERCENT":24, "PERSON":26, "PRODUCT":28, "QUANTITY":30, "TIME":32, "WORK_OF_ART":34}
sep = []
cp = -1
self.entity_indices = []
ets = self.frame['entities']
for x in range(0, len(ets)):
eheld = ets[x][0].split(' ')
sep.append([ets[x][1]])
cp+=1
for y in range(0, len(eheld)):
sep[cp].append(eheld[y])
for y in range(0, len(sep)):
for x in range(0, len(self.track)):
if(self.track[x].qual == 'node'):
totie = []
for z in range(1, len(sep[y])):
if(self.track[x].text == sep[y][z]):
self.track[x].entity_tag = sep[y][0]
totie.append(x)
self.entity_indices.append(x)
if(len(totie)>0):
q = entity_bind()
q.points = totie
for x in range(0, len(totie)):
self.track[x].bind = q
return(self.entity_indices)
#semantic edge, ie connection between two semantic nodes
#carries sentiment charge, negation charge, semantic meaning type, and a weight
#weight is updated by encounter frequency
class sem_edge:
def __init__(self):
self.sentiment_charge = 0.0
self.negation = 0.0
self.type = None
self.weight = 0.0
self.qual = "edge"
class entity_bind:
def __init__(self):
self.points = []
self.qual = "entity_bind"
#vertical traces across time/meaning
class noided:
def __init__(self):
self.end = True
self.qual = "noid"
class sem_trace:
def __init__(self, ax, ay, bx, by):
#starting point
#x is column, y is row
self.ax = ax
self.ay = ay
#ending point
#x is column, y is row
self.bx = bx
self.by = by
#========================
#type (standard, syn, ant, sim, )
#-----------------------
#standard: standard trace
#syn: synonym trace
#ant: antonym trace
#sim: similar trace
#sub: subset of trace
#-----------------------
self.type = "standard"
#========================
#semantic web
class sw:
def __init__(self):
#composed of semantic row adjacency vectors
self.semWeb = []
#composed of semantic column adjacency vectors
self.semTrack = []
#total nodes, for arbitrary querying
self.nodeList = []
self.traces = []
self.relationLabel = []
self.init_special_nodes()
self.e_types = [ "CARDINAL", "DATE", "EVENT", "FAC", "GPE", "LANGUAGE", "LAW", "LOC", "MONEY", "NORP", "ORDINAL", "ORG", "PERCENT", "PERSON", "PRODUCT", "QUANTITY", "TIME", "WORK_OF_ART"]
self.type_lookup = {"CARDINAL":0,"DATE":2,"EVENT":4,"FAC":6, "GPE":8, "LANGUAGE":10, "LAW":12, "LOC":14, "MONEY":16, "NORP":18, "ORDINAL":20, "ORG":22, "PERCENT":24, "PERSON":26, "PRODUCT":28, "QUANTITY":30, "TIME":32, "WORK_OF_ART":34}
self.root_rep = []
def init_special_nodes(self):
types = [ "CARDINAL", "DATE", "EVENT", "FAC", "GPE", "LANGUAGE", "LAW", "LOC", "MONEY", "NORP", "ORDINAL", "ORG", "PERCENT", "PERSON", "PRODUCT", "QUANTITY", "TIME", "WORK_OF_ART"]
types_name = ["numbers", "dates", "events", "facilities", "countries", "language", "laws", "locations", "monetary", "groups", "order", "organization", "percentage", "people", "objects", "measurements", "times", "titles"]
ctrack = []
#make node for each type
for ind in range(0, len(types)):
q = entity_node(types_name[ind], types[ind])
q.node_x = 0
q.node_y = len(self.semWeb)
#insert into nodelist
self.nodeList.append(q)
#insert into track for semvector
ctrack.append(self.nodeList[ind])
#add an edge
co = sem_edge()
ctrack.append(co)
#noid
end = noided()
ctrack.append(end)
#wrap in semvector
forweb = sem_vector()
forweb.text = types_name
forweb.track = ctrack
#insert into semweb
self.semWeb.append(forweb)
def state_insert(self, snf):
#for creation of a state vector
vc = snf.VerbCloud
oc = snf.ObjectCloud
dc = snf.DescriptorCloud
total = len(vc)*len(oc)*len(dc)
for x in range(0, total):
pass
pass
def update_root_rep(self):
bookmark = self.semWeb[self.recent_entry()].track
current_root_rep = []
for x in range(0, len(bookmark)):
if(bookmark[x].qual == "node"):
current_root_rep.append(bookmark[x].alt_form)
self.root_rep.append(current_root_rep)
#print(current_root_rep)
#print(self.root_rep)
def recent_entry(self):
if(len(self.semWeb)==0):
return 0
elif(len(self.semWeb)>=1):
return(len(self.semWeb)-1)
def export_to_json(self):
json_node_form = {}
for row in range(0, len(self.semWeb)):
for nodec in range(0, len(self.semWeb[row].track)):
if(self.semWeb[row].track[nodec].qual == "node" or self.semWeb[row].track[nodec].qual == "entity_node"):
#self.semWeb[row].track[nodec].text
json_slot = str(row)+'-'+str(nodec)
json_node_form[json_slot] = {"text": None, "hash": None}
json_node_form[json_slot]["text"] = self.semWeb[row].track[nodec].text
json_node_form[json_slot]["hash"] = self.semWeb[row].track[nodec].semHash
json_edge_form = {}
for x in range(0, len(self.traces)):
json_edge_form[x] = {"startkey": None, "endkey": None}
json_edge_form[x]["startkey"] = str(self.traces[x].ax) +'-'+str(self.traces[x].ay)
json_edge_form[x]["endkey"] = str(self.traces[x].bx) +'-'+str(self.traces[x].by)
json_node_form['edges'] = json_edge_form
with open('interface/assets/imprisoned_web.json', 'w') as outfile:
json.dump(json_node_form, outfile)
def hash_word_combo(self, wordText, pos_tag):
return(hashlib.md5((wordText+pos_tag).encode()))
#find by hash
def find_web_index_by_hash(self, to_locate_hash):
#O(n) time
indices = []
rev_nodeList = self.nodeList
rev_nodeList.reverse()
for x in range(0, len(rev_nodeList)):
if(rev_nodeList[x].semHash == to_locate_hash):
#print(rev_nodeList[x].text)
indices.append([rev_nodeList[x].node_x, rev_nodeList[x].node_y])
return(indices)
#find by text
def find_web_index_by_text(self, to_locate_text):
#O(n) time
indices = []
rev_nodeList = self.nodeList
rev_nodeList.reverse()
for x in range(0, len(rev_nodeList)):
if(rev_nodeList[x].text == to_locate_text):
indices.append([rev_nodeList[x].node_x, rev_nodeList[x].node_y])
return(indices)
#nodeEncounter, add semnode
def nodeEncounter(self, frame, current):
#get relevant information out of the sentence frame
wordText = frame['plaintext'][current]
pos_tag = frame['tokens'][current][1]
opos_tag = frame['tokens'][current][4]
ents = frame['entities']
#check if text is within an entity
#initialize node index as None
currentNodeIndex = None
#create node hash
tenativeN = hashlib.md5((wordText+pos_tag).encode())
self.nodeList.append(sem_node(wordText, pos_tag, opos_tag))
#attach x,y coordinates to the node
#we haven't appended to the semtrack or semweb so length is correct
currentNodeIndex = len(self.nodeList)
#if length of the nodeList is greater than 1 we can correct for off by 1 error
if(len(self.nodeList)>=1):
currentNodeIndex = currentNodeIndex-1
self.nodeList[currentNodeIndex].node_x = len(self.semTrack)
self.nodeList[currentNodeIndex].node_y = len(self.semWeb)
"""
for eiter in range(0, len(ents)):
if(wordText in ents[eiter][0]):
self.nodeList[currentNodeIndex].entity_tag = ents[eiter][1]
"""
#print(self.nodeList[currentNodeIndex].node_x, self.nodeList[currentNodeIndex].node_y)
#return node index
return(currentNodeIndex, frame['emotional_charge_vector'][current])
def track_construction(self, sentFrame, current, sentcharge):
#toss node into semtrack
self.semTrack.append(self.nodeList[current])
outbound_edge = sem_edge()
if(self.nodeList[current].alt_form=="neg"):
outbound_edge.negation = 1.0
else:
outbound_edge.negation = 0
outbound_edge.sentiment_charge = sentcharge
outbound_edge.type = None
outbound_edge.weight = None
self.semTrack.append(outbound_edge)
def track_stop(self):
q = noided()
self.semTrack.append(q)
#encounter sentence, words into nodes, create
def sentenceEncounter(self, sentFrame):
if(sentFrame == None):
return False
#print(sentFrame['plaintext'])
for x in range(0, len(sentFrame['plaintext'])):
current_nodeXval, sentcharge = self.nodeEncounter(sentFrame, x)
#print(sentFrame['tokens'][x])
if(current_nodeXval!=None):
self.track_construction(sentFrame, current_nodeXval, sentcharge)
#noid track
self.track_stop()
#init vector carrier
vectorized = sem_vector()
#pass track to vector
vectorized.track = self.semTrack
vectorized.frame = sentFrame
vectorized.text = sentFrame['plaintext']
vectorized.text = sentFrame['plaintext']
vectorized.tense = sentFrame['tense']
vectorized.t = dateLib.getNow()
vectorized.speaker = sentFrame['speaker']
#if we have a sentence type prediction, fill it in
if(sentFrame['sent_type_pred']!=None):
vectorized.type = sentFrame['sent_type_pred']
vectorized.entify()
self.correct_list()
#slide in vector to web
self.semWeb.append(vectorized)
self.update_root_rep()
#clear semTrack
self.semTrack = []
#resolve relevancies
#match to 'pool' of relevant information in profiles, states, previous webs
#vertically insert
#find occurence of each node's hash in web
#try iterating upwards until find nearest matching node and then connecting the two
#rather than connecting all at once
#stop connecting stopwords
def spintrace(self):
self.traces = []
for iterator in range(0, len(self.nodeList)):
cx = self.nodeList[iterator].node_x
cy = self.nodeList[iterator].node_y
self.nodeList[iterator].individual_traces = []
self.semWeb[cy].track[cx].individual_traces = []
#print(self.semWeb[cy].track[cx].text)
#print(self.nodeList[iterator].text)
#assert self.semWeb[cy].track[cx].text == self.nodeList[iterator].text
if(self.nodeList[iterator].text in nltk.corpus.stopwords.words('english') or self.nodeList[iterator].text == " "):
pass
else:
totracelist = self.find_web_index_by_hash(self.nodeList[iterator].semHash)
for iterator2 in range(0, len(totracelist)):
self.semWeb[totracelist[iterator2][1]].track[totracelist[iterator2][0]].individual_traces = []
for iterator3 in range(0, len(totracelist)):
if(iterator2!=iterator3 and len(totracelist[iterator2])>1 and len(totracelist[iterator3])>1):
ax = totracelist[iterator2][0]
ay = totracelist[iterator2][1]
bx = totracelist[iterator3][0]
by = totracelist[iterator3][1]
self.traces.append(sem_trace(ax, ay, bx, by))
self.semWeb[by].track[bx].individual_traces.append(sem_trace(bx,by,ax,ay))
self.semWeb[ay].track[ax].individual_traces.append(sem_trace(ax,ay,bx,by))
if(self.nodeList[iterator].entity_tag!='none' and self.nodeList[iterator].qual != 'entity_node'):
targetx = self.type_lookup[self.nodeList[iterator].entity_tag]
#print(targetx)
cnode = self.nodeList[iterator]
#print(self.semWeb[0].track[targetx].entity_tag + ": " + cnode.text)
#print(str(cnode.node_x) + ", " + str(cnode.node_y) + "--->" + str(targetx) + ', 0')
self.semWeb[cnode.node_y].track[cnode.node_x].individual_traces.append(sem_trace(cnode.node_x, cnode.node_y, targetx, 0))
self.semWeb[0].track[targetx].individual_traces.append(sem_trace(targetx, 0, cnode.node_x, cnode.node_y))
self.nodeList[iterator].individual_traces.append(sem_trace(cnode.node_x, cnode.node_y, targetx, 0))
self.traces.append(sem_trace(cnode.node_x,cnode.node_y, targetx, 0))
def aggregate_by_noun_chunks(self, row_index):
print("<--- Retrieving relevant sentences via noun chunks for: -->")
print(" ".join(self.semWeb[row_index].text))
print("<------------------------------------------------------------> ")
chunkindices = self.semWeb[row_index].resolve_chunk_indices()
targeted_nodes = []
aggregatedplaintext = []
for x in range(0, len(chunkindices)):
q = self.semWeb[row_index].track[chunkindices[x]]
for x2 in range(0, len(q.individual_traces)):
if(' ' in self.semWeb[q.individual_traces[x2].by].text):
self.semWeb[q.individual_traces[x2].by].text.remove(' ')
aggregatedplaintext.append(" ".join(self.semWeb[q.individual_traces[x2].by].text))
aggregatedplaintext = ". ".join(aggregatedplaintext)
if(len(aggregatedplaintext)<10):
aggregatedplaintext += " ".join(self.semWeb[row_index].text)
print(aggregatedplaintext)
return(aggregatedplaintext)
#ensure nodeList and semWeb are matching
def correct_list(self):
ncounter = 0
for x in range(0, len(self.nodeList)):
pass
#self.nodeList[x] = self.semWeb[self.nodeList[x].node_y].track[self.nodeList[x].node_x]
def get_by_entity(self, type):
key = self.type_lookup[type]
elist = self.semWeb[0].track[key].individual_traces
#print(key)
#print(elist)
aggregated =[]
for x in range(0, len(elist)):
print(elist[x].by)
aggregated.append(self.semWeb[elist[x].by].track[elist[x].bx].text)
return(aggregated)
def return_track_by_index(self, indx):
return(self.semWeb[indx])
def aggregate_recent_conversation(self):
aggregated = []
return(aggregated)
def aggregate_by_occurence(self, hash):
aggregated =[]
return(aggregated)
def aggregate_by_speaker(self,spkid):
aggregated =[]
return(aggregated)
def tie_node_synonyms(self, node_x, node_y):
syn = wordnet.synsets(self.semWeb[node_y].track[node_x].text)
#print("Word and Type : " + syn[1].name())
#print("Synonym of teaching is: " + syn[1].lemmas()[0].name())
#print("The meaning of the teaching : " + syn[1].definition())
#print("Example of teaching : " + str(syn[1].examples()))
pass
#need a handler to pass different types of statements in, ie add speaker if it was a spoken request
#automatically load and pass in commands
def handler(take_in):
pass
#init vector with nodes,
#graph; time is y, word is X, thus we have a vector of length n, where n is number
#of known words (exposed semantic vocabulary)?
#connecting things between y axis? perhaps subjects, etc
#edges carry several charges across them