Gutenberg Children Books `2019-02-17` MSL=5 `LGParseError with 20 clusters (cell 15)`¶

LG-E-clean corpus, ALE clustering, 2000/1000/500/50/20 clusters, server 94.130.238.118
trash filter off, min_word_count = 1, max_sentence_length' = 5

This notebook is shared as static GCB-LG-E-clean-ALE-MWC=1-MSL=5-2019-02-17_LGParseError.html.
Output data shared via GCB-LG-E-clean-ALE-MWC=1-MSL=5-2019-5-17_LGParseError directory.

Basic settings¶

import os, sys, time
module_path = os.path.abspath(os.path.join('..'))
if module_path not in sys.path: sys.path.append(module_path)
from src.grammar_learner.utl import UTC, test_stats
from src.grammar_learner.read_files import check_dir, check_corpus
from src.grammar_learner.write_files import list2file
from src.grammar_learner.widgets import html_table
from src.grammar_learner.pqa_table import table_rows, params, wide_rows
tmpath = module_path + '/tmp/'
check_dir(tmpath, True, 'none')
start = time.time()
runs = (1,1)
print(UTC(), ':: module_path:', module_path)

2019-02-17 11:16:57 UTC :: module_path: /home/obaskov/94/language-learning

Corpus test settings¶

corpus = 'GCB' # 'Gutenberg-Children-Books-Caps' 
dataset = 'LG-E-clean'
kwargs = {
    'max_sentence_length'   :   5   ,
    'max_unparsed_words'    :   0   ,
    'left_wall'     :   ''          ,
    'period'        :   False       ,
    'context'       :   1           ,
    'min_word_count':   1           ,
    'word_space'    :   'sparse'    ,
    'clustering'    :   ['agglomerative', 'ward'],
    'clustering_metric' : ['silhouette', 'cosine'],
    'cluster_range' :   2000        ,   # 2000/1000/500/50/20
    'top_level'     :   0.01        ,
    'grammar_rules' :   2           ,
    'max_disjuncts' :   1000000     ,   # off
    'stop_words'    :   []          ,
    'tmpath'        :   tmpath      ,
    'verbose'       :   'log+'      ,
    'template_path' :   'poc-turtle',
    'linkage_limit' :   1000        }
rp = module_path + '/data/' + corpus + '/LG-E-clean/GCB-LG-English-clean.ull'
cp = rp  # corpus path = reference_path
runs = (1,1)
out_dir = module_path + '/output/' + 'GCB-LG-E-clean-MWC=1-MSL=5-' + str(UTC())[:10]
print(UTC(), '\n', out_dir)

2019-02-17 11:16:57 UTC 
 /home/obaskov/94/language-learning/output/GCB-LG-E-clean-MWC=1-MSL=5-2019-02-17

Tests: `min_word_count = 1`; 2000/1000/500/50/20 clusters¶

2000/1000/500/50 clusters -- tests passed¶

%%capture
table = []
kwargs['cluster_range'] = 2000
line = [['ALE2000', corpus, dataset, 0, 0, 'none']]
a, _, header, log, rules = wide_rows(line, out_dir, cp, rp, runs, **kwargs)
header[0] = 'Cell'
table.extend(a)

display(html_table([header] + a)); print(test_stats(log))

Cleaned dictionary: 5062 words, grammar learn time: 00:27:40, grammar test time: 00:32:31

%%capture
kwargs['cluster_range'] = 1000
line = [['ALE1000', corpus, dataset, 0, 0, 'none']]
a, _, h, log, rules = wide_rows(line, out_dir, cp, rp, runs, **kwargs)
table.extend(a)

display(html_table([header] + a)); print(test_stats(log))

Cleaned dictionary: 5062 words, grammar learn time: 00:13:02, grammar test time: 00:36:25

%%capture
kwargs['cluster_range'] = 500
line = [['ALE500', corpus, dataset, 0, 0, 'none']]
a, _, header, log, rules = wide_rows(line, out_dir, cp, rp, runs, **kwargs)
header[0] = 'Cell'
table.extend(a)

display(html_table([header] + a)); print(test_stats(log))

Cleaned dictionary: 5062 words, grammar learn time: 00:07:24, grammar test time: 00:40:55

%%capture
kwargs['cluster_range'] = 50
line = [['ALE50', corpus, dataset, 0, 0, 'none']]
a, _, h, log, rules = wide_rows(line, out_dir, cp, rp, runs, **kwargs)
table.extend(a)

display(html_table([header] + a)); print(test_stats(log))

Cleaned dictionary: 5062 words, grammar learn time: 00:02:25, grammar test time: 08:27:38

2000/1000/500/50 clusters tests results¶

display(html_table([header] + table))

print(UTC(), ':: 2000/1000/500/50 finished, elapsed', str(round((time.time()-start)/3600.0, 1)), 'hours')
table_str = list2file(table, out_dir + '/all_tests_table.txt')
print('Results saved to', out_dir + '/all_tests_table.txt')

2019-02-17 22:25:04 UTC :: 2000/1000/500/50 finished, elapsed 11.1 hours
Results saved to /home/obaskov/94/language-learning/output/GCB-LG-E-clean-MWC=1-MSL=5-2019-02-17/all_tests_table.txt

20 clusters test -- failed¶

%%capture
kwargs['cluster_range'] = 20
line = [['ALE20', corpus, dataset, 0, 0, 'none']]
a, _, h, log, rules = wide_rows(line, out_dir, cp, rp, runs, **kwargs)
table.extend(a)

---------------------------------------------------------------------------
LGParseError                              Traceback (most recent call last)
<ipython-input-15-d1ec7b15a3c4> in <module>()
      1 kwargs['cluster_range'] = 20
      2 line = [['ALE20', corpus, dataset, 0, 0, 'none']]
----> 3 a, _, h, log, rules = wide_rows(line, out_dir, cp, rp, runs, **kwargs)
      4 table.extend(a)

~/94/language-learning/src/grammar_learner/pqa_table.py in wide_rows(lines, out_dir, cp, rp, runs, **kwargs)
    427                 for k in range(runs[1]):
    428                     a, f1, precision, q = pqa_meter(re['grammar_file'],
--> 429                                                     og, cp, rp, **kwargs)
    430                     pa.append(a)
    431                     pq.append(q)

~/94/language-learning/src/grammar_learner/pqa_table.py in pqa_meter(dict_path, op, cp, rp, **kwargs)
    112                                              dict_path, grammar_path,
    113                                              template_path, linkage_limit,
--> 114                                              options, reference_path)
    115     return float(pa), float(f1), float(precision), float(recall)
    116 

~/94/language-learning/src/grammar_tester/grammartester.py in test_grammar(corpus_path, output_path, dict_path, grammar_path, template_path, linkage_limit, options, reference_path, timeout)
    372 
    373     # pm, pq = gt.test(dict_path, corpus_path, output_path, reference_path, options, None)
--> 374     pm, pq = gt.test(dict_path, corpus_path, output_path, reference_path, options, TextProgress)
    375 
    376     return \

~/94/language-learning/src/grammar_tester/grammartester.py in test(self, dict_path, corpus_path, output_path, reference_path, options, progress)
    327                 self._options &= (~BIT_DPATH_CREATE)
    328 
--> 329             self._on_dict_file(dict_path, parse_args)
    330 
    331         if self._parser is not None:

~/94/language-learning/src/grammar_tester/grammartester.py in _on_dict_file(self, dict_file_path, args)
    224 
    225         if os.path.isfile(corp_path):
--> 226             self._on_corpus_file(corp_path, [dest_path, lang_path] + args)
    227 
    228         elif os.path.isdir(corp_path):

~/94/language-learning/src/grammar_tester/grammartester.py in _on_corpus_file(self, corpus_file_path, args)
    181 
    182         file_metrics, file_quality = self._parser.parse(dict_path, corpus_file_path, out_file,
--> 183                                                         ref_file, self._options, self._progress)
    184 
    185         if self._options & (BIT_SEP_STAT | BIT_OUTPUT) == BIT_SEP_STAT:

~/94/language-learning/src/grammar_tester/lginprocparser.py in parse(self, dict_path, corpus_path, output_path, ref_file, options, progress)
    298                 # Take an action depending on the output format specified by 'options'
    299                 ret_metrics, ret_quality = self._handle_stream_output(raw.decode("utf-8-sig"), options,
--> 300                                                                       out_stream, ref_file)
    301 
    302                 if progress is not None:

~/94/language-learning/src/grammar_tester/lginprocparser.py in _handle_stream_output(self, text, options, out_stream, ref_path)
    159                     raise LGParseError("Number of sentences in corpus and reference files missmatch. "
    160                                        "Reference file '{}' does not match "
--> 161                                        "its corpus counterpart {} != {}.".format(ref_path, len_ref, len_par))
    162 
    163             sentence_count = 0

LGParseError: Number of sentences in corpus and reference files missmatch. Reference file '/home/obaskov/94/language-learning/data/GCB/LG-E-clean/GCB-LG-English-clean.ull' does not match its corpus counterpart 104341 != 104336.

Test with 20 clusters failed in ~ 102 hours (4d 5h 32m)

Save results¶

display(html_table([header] + table))

print(UTC(), ':: finished, elapsed', str(round((time.time()-start)/3600.0, 1)), 'hours')
table_str = list2file(table, out_dir + '/all_tests_table.txt')
print('Results saved to', out_dir + '/all_tests_table.txt')

2019-02-22 03:57:04 UTC :: finished, elapsed 112.7 hours
Results saved to /home/obaskov/94/language-learning/output/GCB-LG-E-clean-MWC=1-MSL=5-2019-02-17/all_tests_table.txt

Cell	Corpus	Parsing	Space	Linkage	Affinity	G12n	Threshold	Rules	MWC	NN	SI	PA	PQ	F1	Top 5 cluster sizes
ALE2000	GCB	LG-E-clean	cALWEd	ward	euclidean	none	---	2000	1	---	0.0	47%	39%	0.47	[459, 135, 132, 65, 57]
ALE1000	GCB	LG-E-clean	cALWEd	ward	euclidean	none	---	1000	1	---	0.0	50%	39%	0.47	[716, 158, 151, 72, 61]
ALE500	GCB	LG-E-clean	cALWEd	ward	euclidean	none	---	500	1	---	0.0	53%	40%	0.47	[1140, 186, 168, 127, 81]
ALE50	GCB	LG-E-clean	cALWEd	ward	euclidean	none	---	50	1	---	0.0	82%	48%	0.49	[3437, 341, 322, 252, 225]

Gutenberg Children Books 2019-02-17 MSL=5 LGParseError with 20 clusters (cell 15)¶