Open Hub
Open Hub

 Settings |  Report Duplicate

0
I Use This!
Very Low Activity

Commits : Listings

Analyzed about 2 hours ago. based on code collected 1 day ago.
Nov 05, 2025 — Nov 05, 2026
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Section 1638 updated
Srinivasan Kannan
More... about 1 year ago
Updated index.rst
Srinivasan Kannan
More... about 1 year ago
Section 1595 updated
Srinivasan Kannan
More... about 1 year ago
Section 1638 updated
Srinivasan Kannan
More... about 1 year ago
Section 1638 updated
Srinivasan Kannan
More... about 1 year ago
Section 1522 updated
Srinivasan Kannan
More... about 1 year ago
Section 1638 updated
Srinivasan Kannan
More... about 1 year ago
1638. (THEORY and FEATURE) GIS Weather Analytics and Climate Analytics - 17 September 2025 - VARUNA N-Body Climate Model - Navier-Stokes Pressure Gradient computation from user defined average temperature for 4 seasons - N-Body gravity analysis of date range 1/10/2025-1/12/2025 for Sun-Mercury-Venus - related to 148,172,313,445,446,463,1165,1171,1176,1266,1289,1316,1318,1327,1358,1359,1360,1361,1364,1367,1368,1388,1404,1455,1456,1463,1479,1480,1515,1526,1527,1528,1548,1558,1561,1562,1567,1581,1590,1631 and all sections on Correlation between gravitational effects of Mined Celestial N-Body Patterns and terrestrial extreme weather events (or Gravitational effects on climate change), Drone natural obstacle avoidance-Drone autonomous online shopping delivery-Drone Electronic Voting Machine, Artificial Intelligence models for Climate Change-Adaptive Fourier Neural Operator-FourCastNet, Astronomical algorithms, Multifractals, Timeseries analysis, Sequence Mining-Class Association Rules, Global warming,Clausius-Clapeyron equation,Navier-Stokes equation
Srinivasan Kannan
More... about 1 year ago
---------------------------------------------------------------------------------------------------------------------------------------- 1638. (THEORY and FEATURE) GIS Weather Analytics and Climate Analytics - 17 September 2025 - VARUNA N-Body Climate Model - Navier-Stokes Pressure Gradient computation from user defined average temperature for 4 seasons - N-Body gravity analysis of date range 1/10/2025-1/12/2025 for Sun-Mercury-Venus - related to 148,172,313,445,446,463,1165,1171,1176,1266,1289,1316,1318,1327,1358,1359,1360,1361,1364,1367,1368,1388,1404,1455,1456,1463,1479,1480,1515,1526,1527,1528,1548,1558,1561,1562,1567,1581,1590,1631 and all sections on Correlation between gravitational effects of Mined Celestial N-Body Patterns and terrestrial extreme weather events (or Gravitational effects on climate change), Drone natural obstacle avoidance-Drone autonomous online shopping delivery-Drone Electronic Voting Machine, Artificial Intelligence models for Climate Change-Adaptive Fourier Neural Operator-FourCastNet, Astronomical algorithms, Multifractals, Timeseries analysis, Sequence Mining-Class Association Rules, Global warming,Clausius-Clapeyron equation,Navier-Stokes equation ---------------------------------------------------------------------------------------------------------------------------------------- (Disclaimer: Same as Disclaimer in Section 1548)
Srinivasan Kannan
More... about 1 year ago
Section 1226 updated
Srinivasan Kannan
More... about 1 year ago
Section 1623 updated
Srinivasan Kannan
More... about 1 year ago
1637. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 33 - distance relevance of answer to query - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633,1634,1635,1636 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 5 September 2025
Srinivasan Kannan
More... about 1 year ago
-------------------------------------------------------------------------------------------------------------------------------------------- 1637. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 33 - distance relevance of answer to query - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633,1634,1635,1636 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 5 September 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1637.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to import SequenceMatcher string distance computation from difflib. 1637.2 New function most_relevant_sentences() has been implemented and invoked to filter the synthesized sentences and choose only those that are relevant to query as Formal LLM answer. 1637.3 Function def most_relevant_sentences(wikipediaanswer,synthesizedanswer,algorithm="distance",threshold=0.5) at present implements a single clause for levenstein algorithm computation that compares each of the synthesized sentence with wikipedia answer as reference and chooses only those sentences that are of difflib SequenceMatcher ratio less than threshold. Distance measure is the most obvious relevance measure despite being less sophisticated. 1637.4 Formal LLM answers filtered by most_relevant_sentences() are made unique by casting to set(). 1637.5 An example query "Define Agentic AI" is answered by formal LLM as below (from QuestionAnswering.log.5September2025): Formal LLM - Answer to query: ['interpreting the unit of intelligence', 'interpreting this unit , intelligence', 'espionage or political views get valuable secrets', 'espionage or political views get state secrets', 'espionage or political views get your secrets', 'act would result a committee', 'espionage or political views get those secrets', 'interpreting this unit for intelligence', 'interpreting the unit with intelligence', 'act would result your committee', 'espionage or political views get nuclear secrets', 'interpreting this unit and intelligence', 'interpreting this unit with intelligence', 'act would result , committee', 'espionage or political views get trade secrets', 'interpreting this unit of intelligence', 'act would result of committee', 'espionage or political views get some secrets', 'act would result The committee', 'espionage or political views get the secrets', 'act would result our committee', 'interpreting this unit in intelligence', 'act would result this committee', 'interpreting this unit on intelligence', 'act would result the committee', "interpreting this unit 's intelligence", 'espionage or political views get these secrets', 'espionage or political views get their secrets', 'espionage or political views get our secrets', 'act would result that committee', 'interpreting the unit on intelligence', 'act would result from committee', 'act would result in committee'] 1637.6 It is hereby cautioned that earlier answer is partial only because comprehensive computation is infeasible due to limited resources. Corresponding wikipedia answer used as distance reference for earlier query is (both signify espionage or military intelligence as against query on Agentic AI which are autonomous AI agents without human intervention): wssummarysentences: ['Espionage, spying, or intelligence gathering, as a subfield of the intelligence field, is the act of obtaining secret or confidential information (intelligence)', ' A person who commits espionage on a mission-specific contract is called an espionage agent or spy', ' A person who commits espionage as a fully employed officer of a government is called an intelligence officer', ' Any individual or spy ring (a cooperating group of spies), in the service of a government, company, criminal organization, or independent operation, can commit espionage', ' The practice is clandestine, as it is by definition unwelcome', ' In some circumstances, it may be a legal tool of law enforcement and in others, it may be illegal and punishable by law', '\nEspionage is often part of an institutional effort by a government or commercial concern', ' However, the term tends to be associated with state spying on potential or actual enemies for military purposes', ' Spying involving corporations is known as corporate espionage', '\n\nOne way to gather data and information about a targeted organization is by infiltrating its ranks', ' Spies can then return information such as the size and strength of enemy forces', ' They can also find dissidents within the organization and influence them to provide further information or to defect', ' In times of crisis, spies steal technology and sabotage the enemy in various ways', ' Counterintelligence is the practice of thwarting enemy espionage and intelligence-gathering', ' Almost all sovereign states have strict laws concerning espionage, including those who practice espionage in other countries, and the penalties for being caught are often severe', ''] 1637.7 In ngram sentence synthesis an example randomwalk (a,b,c,d) over textgraph is synthesized as sentence: a-ngram1-b-ngram2-c-ngram3-d by sourcing ngrams from ngrams.dev which is a k-order Markov model of text restoration but not statistical because no Bayesian priors are assumed and all possible ngrams are exhaustively iterated. In addition, all the ngram sentences are checked for Part-of-Speech correctness against Penn Treebank grammar trees. 1637.8 Following table illustrates the distinction between Sentence template synthesis and Ngram sentence synthesis (directions are mutually opposite): ------------------------------------------------------------------------------------------------------------------------ Sentence template synthesis ------- Ngrams sentence synthesis ------------------------------------------------------------------------------------------------------------------------ (*) First get inorder traversed (*) First connect the randomwalks of textgraphs by sentence PoS templates from Penn Treebank all possible ngrams pairwise or wildcard (*) Then fill the sentence PoS templates from (*) Then verify the synthesized sentences for Part-of-speech randomwalks of textgraphs compliance with Penn Treebank inorder traversed PoS templates (*) There could be semifilled sentence templates (*) No blanks in the synthesized sentences that require fill-in-the-blanks wildcard search
Srinivasan Kannan
More... about 1 year ago
Section 1255 updated
Srinivasan Kannan
More... about 1 year ago
+1636. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 32 - Word duplication removal in synthesized sentences - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633,1634,1635 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 4 September 2025
Srinivasan Kannan
More... about 1 year ago
-------------------------------------------------------------------------------------------------------------------------------------------- 1636. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 32 - Word duplication removal in synthesized sentences - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633,1634,1635 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 4 September 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1636.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to invoke newly defined remove_duplication() function to rectify ngram sentences that contain duplicate words consecutively. 1636.2 Function def remove_duplication(sentence) scans an ngram synthesized sentence and removes duplicate words that are caused when ngramshop=1 and null tokens found in quick succession returning a new sentence with unique words. Using set() operator to remove duplicates won't work because set() perturbs the part-of-speech sequence in synthesized sentence. 1636.3 Function def verify_part_of_speech(ngramproductsentence,algorithm="topologicalsort",arraytemplates=None) has been updated to refactor topological sort clause to concatenate tokens in topological sort of dependency graph of ngram sentence into another sentential form that can be string-compared with the rest for distinguishing good and bad part-of-speech sequences. 1636.4 Number of inorder traversed grammar trees has been increased by setting to default: list_of_sentence_PoS_arrays=create_sentence_PoS_dict_from_treebank(returnasarray=True) 1636.5 WikipediaRLFGTransformersQuestionAnswering() is invoked with ngramshop=1 and number_of_cores_per_random_walk=1 parameters to increase probability of meaningful sentence synthesis (and the caveat as usual: this implementation requires huge computation to synthesize meaningful sentences from all possible random walks in a comprehensive textgraph behemoth): WikipediaRLFGTransformersQuestionAnswering(question,wsheading=False,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=5,std_sentence_PoS_dict={},number_of_cores_per_random_walk=1,number_of_random_walks=3,blanks=False,treenode_type="tag",sentence_tuple_array=True, sentence_PoS_array=sentencePoSarray, randomwalk_to_sentence_template_ratio=3,user_defined_PoS2Vocabulary_dict=None,ngrams_sentence_synthesis_randomwalks=True,ngrams_sentence_synthesis_fillintheblanks=False,pairwise_ngram=True,ngramshop=1) 1636.6 Two queries posed to LLM have been answered as below in logs without duplicate words: Query1: "What is the difference between C and C++" - testlogs/QuestionAnswering.log.4September2025 Query2: "Define Finite State Automaton" - testlogs/QuestionAnswering.log2.4September2025 1636.7 One of the ngram sentences synthesized was queried in DeepSeek LLM for perplexity score - chat transcript from https://chat.deepseek.com/a/chat/s/ac59e862-502d-4f31-bed2-a8f84f0ab165 below - GPT-2 transformers model: from transformers import GPT2LMHeadModel, GPT2TokenizerFast import torch # Load model and tokenizer model_id = "gpt2" model = GPT2LMHeadModel.from_pretrained(model_id) tokenizer = GPT2TokenizerFast.from_pretrained(model_id) # Define the sentence sentence = "one western state welfare system without government in transition" # Tokenize and prepare inputs inputs = tokenizer(sentence, return_tensors='pt') with torch.no_grad(): outputs = model(**inputs, labels=inputs['input_ids']) # Calculate loss and perplexity loss = outputs.loss perplexity = torch.exp(loss).item() print(f"Sentence: '{sentence}'") print(f"Perplexity: {perplexity:.2f}")
Srinivasan Kannan
More... about 1 year ago
Section 1630 updated
Srinivasan Kannan
More... about 1 year ago
Section 1635 updated
Srinivasan Kannan
More... about 1 year ago
1635. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 31 - Part-of-speech verification by exhaustive grammar tree search and dependency graph topological sort - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633,1634 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 31 August 2025
Srinivasan Kannan
More... about 1 year ago
-------------------------------------------------------------------------------------------------------------------------------------------- 1635. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 31 - Part-of-speech verification by exhaustive grammar tree search and dependency graph topological sort - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633,1634 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 31 August 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1635.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to import NetworkX. Number ngram sentences synthesized is limited by max_synth_sentences parameter. 1635.2 New function verify_part_of_speech(ngramproductsentence,algorithm) has been implemented and invoked to verify part-of-speech of ngram sentences synthesized by two algorithms chosen by algorithm="topologicalsort" or algorithm="exhaustivesearch" options: (1) Topological sort of DisplaCy dependency graph of sentences (2) Exhaustive search of PoS sequence of sentences in inorder traversed PoS sequences of Penn Treebank grammar trees (which might succeed very rarely due to size of corpus that needs to be searched) 1635.3 Function def verify_part_of_speech(ngramproductsentence,algorithm="toplogicalsort",arraytemplates=None) imports SpaCy and DisplaCy and computes the PoS sequence of ngram synthesized sentence from SpaCy en_core_web_sm model. DisplaCy.parse_deps() code has been refactored into this function to instantiate a DiGraph NetworkX object from start and end attributes of arcs in DisplaCy dependency graph. 1635.4 First if clause computes topological sort of earlier dependency DiGraph object prints the sorted vertices. Second if clause implements the exhaustive bruteforce search of PoS sequence of ngram sentence within a user-supplied array of array sentence templates parsed from inorder traversals of Penn Treebank grammar trees. This exhaustive search may not always succeed as outcome depends on number of array sentence templates supplied which could be in millions sometimes - search is considered successful if ngram PoS sequence is a subsequence within grammar tree PoS sequence ("in" operator). If none of the sentence templates have a matching PoS, result is printed as "Invalid PoS" and loop breaks with "Valid PoS" message if a matching PoS is found. 1635.5 Three example sentences are verified for PoS as below with topologicalsort and exhaustivesearch algorithms: verify_part_of_speech("All you need is attention",algorithm="toplogicalsort") verify_part_of_speech("All you need is attention",algorithm="exhaustivesearch",arraytemplates=list_of_sentence_PoS_arrays) verify_part_of_speech("Attention is all you need",algorithm="toplogicalsort") verify_part_of_speech("is attention all you need",algorithm="exhaustivesearch",arraytemplates=list_of_sentence_PoS_arrays) 1635.6 An example Drug discovery query for Attention-Deficit-Hyperactivity-Disorder and CNS stimulant medicine Methamphetamine: "Medicinal uses of Methamphetamine" is answered by some of the following ngram sentences in QuestionAnswering.log.31August2025 along with PoS verifications earlier (Topological sorts and Exhaustive search - Distinguishing topological sorts for Good and Bad PoS sequences is still an unknown): ... WordNet-ConceptNet TextGraph Random Walk - Bot generated answers ranked by wordnet perplexity: {0.0006581653640477172: ['nervous system relating recreation or recreational '], 0.0008044243338360986: ['nervous mechanisms relating recreation or recreational '], ...
Srinivasan Kannan
More... about 1 year ago
Section 1630 updated
Srinivasan Kannan
More... about 1 year ago
added testlogs/CUMTA_CMAPopulationEstimates2025.jpeg
Srinivasan Kannan
More... about 1 year ago
Updated Requirements.txt
Srinivasan Kannan
More... about 1 year ago
Section 1634 updated
Srinivasan Kannan
More... about 1 year ago
Section 1634 updated
Srinivasan Kannan
More... about 1 year ago
1634. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 30 - Additional boolean flag for clauses, Spacy-Displacy Dependencies and Perplexity rankings - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 24 August 2025
Srinivasan Kannan
More... about 1 year ago
-------------------------------------------------------------------------------------------------------------------------------------------- 1634. (THEORY and FEATURE) Nachiketas Formal LLM - Google Ngrams V3 dataset (Ngrams.dev) sentence synthesizer - Changes to Wikipedia Question-Answering algorithm for better accuracy 30 - Additional boolean flag for clauses, Spacy-Displacy Dependencies and Perplexity rankings - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578,1579,1583,1585,1587,1595,1611,1620,1621,1632,1633 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression, Drug discovery, Universal Chomsky Sentence Synthesizer, Formal AI, Perplexity in Statistical AI - 24 August 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1634.1 Path from pathlib has been imported. Function def WikipediaRLFGTransformersQuestionAnswering(question,questionfraction=1,maxanswers=1,keywordsearch=False,wsheading=True,answerslice=1,answerfraction=1,bothvertices_intersection=True,sentence_type="xtag_node34_triplets",number_of_random_walks=10,number_of_words_per_sentence=5,number_of_cores_per_random_walk=5,std_sentence_PoS_dict={"ADJ":[],"PROPN":[],"NOUN":[],"AUX":[],"ADP":[],"ADV":[],"VERB":[],"DET":[],"PRON":[],"CCONJ":[],"NUM":[],"SYM":[],"X":[]},blanks=False,perplexity_algorithm="WordNet",treenode_type="PoS",sentence_tuple_array=False,sentence_PoS_array=None,randomwalk_to_sentence_template_ratio=10,user_defined_PoS2Vocabulary_dict=None,transformers_enabled=False,ngrams_sentence_synthesis_randomwalks=True,ngrams_sentence_synthesis_fillintheblanks=False,pairwise_ngram=False,ngramshop=2) has been enhanced to include boolean flags ngrams_sentence_synthesis_randomwalks, ngrams_sentence_synthesis_fillintheblanks and ngramshop that respectively enable (*) randomwalk ngram sentence synthesis (*) wildcard sentence template ngram synthesis and (*) configure the number of hops in pairwise ngrams - valued 1 (overlapping ngrams) or 2 (non-overlapping ngrams) and the flags are propagated to make_sentence2(). 1634.2 From spacy, displacy visualizer is imported. Each of the ngram synthesized sentences is parsed for dependencies by displacy.parse_deps() which is a graph representation of a sentence that can be rendered visually as SVG file (commented at present). Displacy Dependency Visualizer implements (Head-Dependents) Dependency Grammars described in : (*) Speech and Language Processing - [Daniel Jurafsky & James H. Martin] - https://web.stanford.edu/~jurafsky/slp3/old_oct19/15.pdf (*) Dependencies and Dependency Grammar - Section 5 - https://www.nltk.org/book/ch08.html (*) Compositional semantics of Feature based grammar - Section 4.1 - https://www.nltk.org/book/ch10.html and graphs for each ngram synthesized sentence are written as a dictionary. Dependency parsing of ngram synthesized sentences has been implemented as a PoS-verification algorithm at present as a simpler alternative though ideally PoS-compliance has to be ensured by extracting PoS sequence of a sentence and checking if the sequence matches against atleast one of the inorder PoS traversals of entire set of Penn Treebank grammars which is exponentially hard. 1634.3 Ngram synthesized sentences are ranked by perplexity measures (pyplexity or WordNet) that could be a basis of selecting meaningful sentences. 1634.4 Ngram sentence synthesis is carried out as follows for pairwise ngrams over randomwalks with ngramshop=2 : WikipediaRLFGTransformersQuestionAnswering(question,wsheading=False,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=5,std_sentence_PoS_dict={},number_of_cores_per_random_walk=3,number_of_random_walks=3,blanks=False,treenode_type="tag",sentence_tuple_array=True, sentence_PoS_array=sentencePoSarray, randomwalk_to_sentence_template_ratio=3,user_defined_PoS2Vocabulary_dict=None,ngrams_sentence_synthesis_randomwalks=True,ngrams_sentence_synthesis_fillintheblanks=False,pairwise_ngram=True,ngramshop=2) 1634.5 Following is an answer ngram sentence with its WordNet perplexity score and its displacy dependency parsing head-dependents graph for query "What is the sum of one and two" and the answer is not "three" but based on topological meaning of sum: "Connected sum of two manifolds in topology" - from testlogs/QuestionAnswering.log.24August2025 : 0.00021626297577854672: ['performance or activity process or operation '], .... ngram synthesized sentence - dependencies: {'words': [{'text': 'performance', 'tag': 'NOUN', 'lemma': None}, {'text': 'or', 'tag': 'CCONJ', 'lemma': None}, {'text': 'activity', 'tag': 'NOUN', 'lemma': None}, {'text': ' ', 'tag': 'SPACE', 'lemma': None}, {'text': 'process', 'tag': 'NOUN', 'lemma': None}, {'text': 'or', 'tag': 'CCONJ', 'lemma': None}, {'text': 'operation', 'tag': 'NOUN', 'lemma': None}], 'arcs': [{'start': 0, 'end': 1, 'label': 'cc', 'dir': 'right'}, {'start': 2, 'end': 4, 'label': 'compound', 'dir': 'left'}, {'start': 2, 'end': 3, 'label': 'dep', 'dir': 'right'}, {'start': 0, 'end': 4, 'label': 'conj', 'dir': 'right'}, {'start': 4, 'end': 5, 'label': 'cc', 'dir': 'right'}, {'start': 4, 'end': 6, 'label': 'conj', 'dir': 'right'}], 'settings': {'lang': 'en', 'direction': 'ltr'}}
Srinivasan Kannan
More... about 1 year ago
Section 1630 updated
Srinivasan Kannan
More... about 1 year ago
Section 1584 updated
Srinivasan Kannan
More... about 1 year ago
Section 1633 updated
Srinivasan Kannan
More... about 1 year ago