| Google Maps Datadriven Styling Urban Area Boundary Statistics for 2 Urban Areas |
|
More...
|
almost 2 years ago
|
| Section 1576 updated |
|
More...
|
almost 2 years ago
|
| Erroneous street name corrected manually |
|
More...
|
almost 2 years ago
|
| 1577. (THEORY and FEATURE) GIS Live Traffic and Weather analytics - METAR live weather data from airport code - 20 January 2025 - related to 823,1163,1164,1174,1175,1177,1179,1194,1211,1226,1234,1290,1291,1292,1295,1296,1297,1301,1302,1304,1312,1313,1320,1323,1326,1328,1342,1343,1349,1350,1351,1362,1363,1377,1386,1387,1395,1397,1398,1412,1414,1423,1448,1449,1462,1471,1472,1476,1529,1531,1545,1546 and all sections on GIS Urban Sprawl Analytics, Merit of Large Scale Visuals, Gridded Population Estimators (SEDAC GPW4 and GHSL convex hull estimators), DBSCAN and Neural Network Urban Extent Learners, Computational Geometry, Intrinsic Merit Rankings, Verlhulste-R Ricker Population Estimation, Community detection, Social Network Analytics, Space filling by Curves, Unequal Circle and Polynomial packing, Computational Geometry and Polygonal packing by Voronoi Tessellation, Four color theorem, Economic Merit, Computational Economics - GDP estimation from Urban sprawl demographics, electricity consumption, Poverty alleviation, Preferential Attachment, Automatic Delineation of Urban Sprawls, Cellular Automata-SLEUTH-Polya Urn-Dynamic Facegraph Urban Growth Models,Expander Graphs,Timeseries analysis,Graph Edit Distance and Earth mover distance, Face recognition,Graph Isomorphism (Exact, Approximate and Subgraph),Graph matching,Graph similarity,Image Search,Quadrilateral Mesh Deformation and Isomorphism,Maximum Common Subgraph Problem and Approximate Topological Matching,TomTom and HERE live road traffic API,Autonomous Online Shopping Delivery by Cargo drones,EventNet GEM |
|
More...
|
almost 2 years ago
|
| -------------------------------------------------------------------------------------------------------------------------------- 1577. (THEORY and FEATURE) GIS Live Traffic and Weather analytics - METAR live weather data from airport code - 20 January 2025 - related to 823,1163,1164,1174,1175,1177,1179,1194,1211,1226,1234,1290,1291,1292,1295,1296,1297,1301,1302,1304,1312,1313,1320,1323,1326,1328,1342,1343,1349,1350,1351,1362,1363,1377,1386,1387,1395,1397,1398,1412,1414,1423,1448,1449,1462,1471,1472,1476,1529,1531,1545,1546 and all sections on GIS Urban Sprawl Analytics, Merit of Large Scale Visuals, Gridded Population Estimators (SEDAC GPW4 and GHSL convex hull estimators), DBSCAN and Neural Network Urban Extent Learners, Computational Geometry, Intrinsic Merit Rankings, Verlhulste-R Ricker Population Estimation, Community detection, Social Network Analytics, Space filling by Curves, Unequal Circle and Polynomial packing, Computational Geometry and Polygonal packing by Voronoi Tessellation, Four color theorem, Economic Merit, Computational Economics - GDP estimation from Urban sprawl demographics, electricity consumption, Poverty alleviation, Preferential Attachment, Automatic Delineation of Urban Sprawls, Cellular Automata-SLEUTH-Polya Urn-Dynamic Facegraph Urban Growth Models,Expander Graphs,Timeseries analysis,Graph Edit Distance and Earth mover distance, Face recognition,Graph Isomorphism (Exact, Approximate and Subgraph),Graph matching,Graph similarity,Image Search,Quadrilateral Mesh Deformation and Isomorphism,Maximum Common Subgraph Problem and Approximate Topological Matching,TomTom and HERE live road traffic API,Autonomous Online Shopping Delivery by Cargo drones,EventNet GEM -------------------------------------------------------------------------------------------------------------------------------- 1577.1 python-src/image_pattern_mining/ImageNet/GISLiveTraffic.py has been updated to enhance function def flightradar24_live_air_traffic(zones=None,max_no_of_flights=10,radius=1000,latitude=12.439259,longitude=79.271851,airport_code="MAA") by adding arguments for longitude-latitude, airport code and radius around airport for fetching live flights data. 1577.2 FlightRadar24 Code commented earlier for fetching live flight data within a circular region centered around a longitude-latitude has been uncommented and is executed based on additional location and radius arguments earlier. 1577.3 New function get_metar_weather_data(airport_code) has been defined which gets live METAR (METeorological Aerodrome Report) weather data for a city airport code and returns it - this weather data includes current wind speed, temperature and pressure millibar data for a city that could find use in GISWeatherAnalytics. 1577.4 Logs obtained by venv/bin/python3.12 GISLiveTraffic.py 2>&1 > testlogs/GISLiveTraffic.log.20January2025 reflect earlier code changes and print METAR weather data for a city airport code. 1577.5 Airport Temporal and Current Flow Betweenness Centrality measures have been recomputed for live air traffic data ranking them by importance - excerpts below from testlogs/GISLiveTraffic.log.20January2025: Temporal Betweenness Centrality of Air Traffic Dynamic Graph: [('SCL', 0.00429806464289223), ('MEX', 0.0037884369780921507), ('IST', 0.00206307102858827), ('LAX', 0.0020078101974653696), ('SIN', 0.0016209843796050692), ('DOH', 0.0015473032714412023), ('MLE', 0.0015473032714412023), ('DFW', 0.0011052166224580018), ('MAD', 0.0010745161607230572), ('LHR', 0.001031535514294135), ('HEL', 0.0009578544061302681), ('EWR', 0.0009578544061302681), ('ATL', 0.0007920719127615679), ('LIM', 0.00066927006582179), ('UIO', 0.00048506729541212304), ('IAH', 0.0004420866489832007), ('MEL', 0.0004420866489832007), ('FRA', 0.0004420866489832007), ('JFK', 0.0004420866489832007), ('JNB', 0.0004420866489832007), ('TLV', 0.0003684055408193339), ('DXB', 0.0003684055408193339), ('ADD', 0.00022104332449160034), ('IAD', 0.00014736221632773356), ('ANC', 0.00014736221632773356), ('BFI', 0.00014736221632773356), ('BOG', 0.00014736221632773356), ('MUC', 7.368110816386678e-05), ('WAW', 7.368110816386678e-05), ('TRN', 0.0), ('ISL', 0.0), ('TEB', 0.0), ('ZRH', 0.0), ('RDU', 0.0), ('CDG', 0.0), ('CAY', 0.0), ('ORY', 0.0), ('CPT', 0.0), ('AZI', 0.0), ('LBG', 0.0), ('CLT', 0.0), ('OPF', 0.0), ('CMH', 0.0), ('RNO', 0.0), ('GUF', 0.0), ('VNY', 0.0), ('BUF', 0.0), ('BNA', 0.0), ('LRD', 0.0), ('MHR', 0.0), ('IAG', 0.0), ('SJU', 0.0), ('SDF', 0.0), ('CPH', 0.0), ('MIA', 0.0), ('FLL', 0.0), ('SAP', 0.0), ('XPL', 0.0), ('EZE', 0.0), ('SAL', 0.0), ('CTU', 0.0), ('PER', 0.0), ('SYD', 0.0), ('BKK', 0.0), ('DAC', 0.0), ('BNE', 0.0), ('PNH', 0.0), ('HKG', 0.0), ('CNS', 0.0), ('TFU', 0.0), ('KUL', 0.0), ('PVG', 0.0), ('DMK', 0.0), ('KIX', 0.0), ('SGN', 0.0), ('TPE', 0.0), ('CGK', 0.0), ('HND', 0.0), ('NRT', 0.0), ('HAN', 0.0), ('CTS', 0.0), ('PEK', 0.0), ('TAS', 0.0), ('ICN', 0.0), ('URC', 0.0), ('SYX', 0.0), ('MRU', 0.0), ('MED', 0.0), ('GRU', 0.0), ('LYS', 0.0), ('FCO', 0.0), ('BEY', 0.0), ('NBO', 0.0), ('HOP', 0.0), ('HHN', 0.0), ('LOS', 0.0), ('MCO', 0.0), ('MAN', 0.0), ('LFW', 0.0), ('MDE', 0.0), ('ACC', 0.0), ('HAV', 0.0), ('LIS', 0.0), ('SDQ', 0.0), ('VIE', 0.0), ('DDD', 0.0), ('GYD', 0.0), ('GKK', 0.0), ('ATH', 0.0), ('CUN', 0.0), ('YYC', 0.0), ('KEF', 0.0), ('BER', 0.0), ('DEL', 0.0), ('SLC', 0.0), ('AMS', 0.0), ('SFO', 0.0), ('YYZ', 0.0)] Current flow betweenness centrality exception: Graph not connected. Approximate Current flow betweenness centrality exception: Graph not connected. Edge Current flow betweenness centrality exception: Graph not connected. |
|
More...
|
almost 2 years ago
|
| Section 1576 updated |
|
More...
|
almost 2 years ago
|
| Section 1576 updated |
|
More...
|
almost 2 years ago
|
| Section 1576 updated |
|
More...
|
almost 2 years ago
|
| Section 1576 updated |
|
More...
|
almost 2 years ago
|
| Section 1576 updated |
|
More...
|
almost 2 years ago
|
| 1576. (THEORY and FEATURE) Nachiketas Formal LLM - Semifilled sentence templates have been filled to synthesize grammatically correct Chomsky sentences - Changes to Wikipedia Question-Answering algorithm for better accuracy 15 - 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 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 - 12 January 2025 |
|
More...
|
almost 2 years ago
|
| -------------------------------------------------------------------------------------------------------------------------------------------- 1576. (THEORY and FEATURE) Nachiketas Formal LLM - Semifilled sentence templates have been filled to synthesize grammatically correct Chomsky sentences - Changes to Wikipedia Question-Answering algorithm for better accuracy 15 - 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 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 - 12 January 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1576.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to import ConditionalFreqDist from nltk. 1576.2 Function def make_sentence2(randomwalkvertices,sentence_PoS_tuple_array=[],treenode_type="tag",max_synth_sentences=1000) has been extended to include an argument for limiting maximum number of synthesized sentences. 1576.3 ConditionalFreqDist object is instantiated from Penn Treebank tagged_words Wall Street Journal corpus and set of words corresponding to a Penn Treebank PoS tag name are obtained from ConditionalFreqDist. 1576.4 Set of semifilled sentences, obtained earlier by pigeonhole-ing textgraph randomwalks onto grammar tree inorder traversal PoS templates,are iterated and PoS tag names in semifilled sentence templates are replaced by all possible words for those PoS names. 1576.5 Earlier process generates a 2-Dimensional array which is flattened by itertools.product() to get cartesian product of all possible fully filled Chomsky sentences which are grammatically correct superset of meaningful sentences. 1576.6 Set of meaningful sentences is the subset of earlier Chomsky sentences synthesized which are ranked by wordnet perplexity scores. New Penn Treebank dataset wsj_0005.mrg has been used for creating inorder traversed sentence templates from grammar trees and query "define odd integers" is posed to the LLM and answer sentences which are fully filled have been logged to QuestionAnswering.log.12January2025. 1576.7 Earlier code change attains a crucial milestone of getting the grammar right for sentences synthesized which have to be further sieved to filter out less meaningful sentences - Apart from perplexity another allternative to weigh the meaning of a grammatically correct synthesized sentence is to define textgraph for it and compute intrinsic merit by graph complexity measures. |
|
More...
|
almost 2 years ago
|
| Section 1560 updated |
|
More...
|
almost 2 years ago
|
| Section 1574 updated |
|
More...
|
almost 2 years ago
|
| Section 1574 updated |
|
More...
|
almost 2 years ago
|
| Section 1575 updated |
|
More...
|
almost 2 years ago
|
| Section 1575 updated |
|
More...
|
almost 2 years ago
|
| Section 1476 updated |
|
More...
|
almost 2 years ago
|
| Section 1215 updated |
|
More...
|
almost 2 years ago
|
| 1575. (THEORY and FEATURE) Nachiketas Formal LLM - Randomwalk to sentence template lengths ratio and more Penn Treebank datasets added - Changes to Wikipedia Question-Answering algorithm for better accuracy 14 - 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 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 - 8 January 2025 |
|
More...
|
almost 2 years ago
|
| -------------------------------------------------------------------------------------------------------------------------------------------- 1575. (THEORY and FEATURE) Nachiketas Formal LLM - Randomwalk to sentence template lengths ratio and more Penn Treebank datasets added - Changes to Wikipedia Question-Answering algorithm for better accuracy 14 - 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 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 - 8 January 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1575.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to enhance 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) to add additional argument randomwalk_to_sentence_template_ratio which is the ratio of lengths of textgraph-knowledgegraph randomwalk and sentence PoS tag template. This ratio is crucial in minimizing the number of unfilled PoS slots in synthesized sentences thus increasing accuracy by making the lengths of textgraph randomwalks and sentence templates proportional - theoretically lower ratios should be more accurate. 1575.2 Clause for checking above ratio has been introduced and make_sentence2() is invoked predicated by this if clause. 1575.3 More Penn Treebank-3 grammar tree datasets has been added to Question-Answering function which is invoked for randomwalk_to_sentence_template_ratio=1 (that is randomwalks can be of same size as sentence PoS template from grammar tree inorder traversals so that pigeonhole process of one-one PoS-word mapping is near-exact): WikipediaRLFGTransformersQuestionAnswering(question,wsheading=True,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=50,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=1) 1575.4 More information in logs for above code changes (3.4 GB gzipped to 40 MB): testlogs/QuestionAnswering.log.8January2025.gz 1575.5 Any graph sampling metric other than textgraph randomwalks could be chosen for filling the sentence templates but choice of randomwalks achieves 2 objectives: (*) randomwalks are random samples of the textgraph word vertices and (*) randomwalks semantically connect the vertices along the walks thereby enshrining a meaning to the walks apriori. |
|
More...
|
almost 2 years ago
|
| Section 1476 updated |
|
More...
|
almost 2 years ago
|
| 1574. (THEORY and FEATURE) Nachiketas Formal LLM - Optional clause for array format of grammar tree inorder traversal sentence template - Changes to Wikipedia Question-Answering algorithm for better accuracy 13 - 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 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 - 7 January 2025 |
|
More...
|
almost 2 years ago
|
| -------------------------------------------------------------------------------------------------------------------------------------------- 1574. (THEORY and FEATURE) Nachiketas Formal LLM - Optional clause for array format of grammar tree inorder traversal sentence template - Changes to Wikipedia Question-Answering algorithm for better accuracy 13 - 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 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 - 7 January 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1574.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to enhance 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) to add arguments sentence_tuple_array (boolean) and sentence_POS_array (array of grammar tree nodes flattened by inorder traversal) which is the sentence template in array format. 1574.2 Based on if sentence_tuple_array is True or False either make_sentence2() or make_sentence() are invoked - make_sentence2() is the new function defined for filling grammar tree inorder traversal sentence template array from textgraph randomwalks. Array sentence template is necessary to allow multimap behaviour having duplicate keys which is not possible in a dictionary e.g Two NNPs in same template is possible in arrays while dictionary template can have only a unique NNP tag. 1574.3 New function def make_sentence2(randomwalkvertices,sentence_PoS_tuple_array=[],treenode_type="tag") has been defined to iterate through a textgraph random walk and fill sentence_PoS_tuple_array (array format sentence template instead of dictionary template in make_sentence()) from SpaCy PoS tagged textgraph randomwalk tokens. Unfilled slots in array sentence template are marked with their grammar tree tag names. Itertools product is then invoked to get cartesian product of all possible sentences (which can be incomplete due to unfilled PoS slots). 1574.4 Function def create_sentence_PoS_dict_from_treebank(datasets=["wsj_0001.mrg"],returnasarray=True) has been updated to read array of Penn Treebank grammar tree datasets and a boolean argument returnasarray to return sentence template as array has been added. 1574.5 WikipediaRLFGTransformersQuestionAnswering() has been benchmarked again with above changes for array sentence templates inferred from Penn Treebank: list_of_sentence_PoS_arrays=create_sentence_PoS_dict_from_treebank(datasets=["wsj_0001.mrg"],returnasarray=True) for sentencePoSarray in list_of_sentence_PoS_arrays: print("sentencePoSarray:",sentencePoSarray) WikipediaRLFGTransformersQuestionAnswering(question,wsheading=True,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=50,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) 1574.6 In summary following is the Nachiketas Formal LLM algorithm implemented so far: 1574.6.1 Read a corpus of grammar annotated Penn Treebank sentences 1574.6.2 Extract the grammar trees from Penn Treebank Wall Street Journal sentence corpus 1574.6.3 Flatten the grammar trees by inorder traversal which yields set of sentence templates in dictionary or array format mapping <Part-of-Speech> to <array of textgraph randomwalk vertices matching the Part-of-speech> 1574.6.4 Iterate through set of sentence templates inferred from Penn Treebank and try to fill the templates to the maximum from textgraph randomwalk SpaCy PoS tagged vertices - unfilled PoS slots are marked with their tag names 1574.6.5 Above process results in sentence templates which are either fully filled or partially filled with words 1574.6.6 Compute cartesian product of earlier semi-filled or filled sentence templates which are close to human sentences. 1574.6.6 Unfilled PoS tag names in sentence templates could be completed by choosing a permutation of words matching the unfilled PoS tag name from a corpus. 1574.6.7 Accuracy of above algorithm (i.e reduction in number of unfilled PoS slots in the template) increases with number of sentence templates extracted from Penn Treebank corpus and number of textgraph randomwalks traversed - in other words complexity bound for above algorithm is approximately O(number_of_sentence_templates * number_of_textgraph_randomwalks) which is massive for large textgraphs of million randomwalks and large corpus of million sentence templates implying synthesis of even a page of meaningful sentences could be daunting. 1574.6.8 Penn Treebank-3 of 1 million words is described in LDC as - https://catalog.ldc.upenn.edu/LDC99T42 :- ".... The Penn Treebank (PTB) project selected 2,499 stories from a three year Wall Street Journal (WSJ) collection of 98,732 stories for syntactic annotation. These 2,499 stories have been distributed in both Treebank-2 (LDC95T7) and Treebank-3 (LDC99T42) releases of PTB. Treebank-2 includes the raw text for each story. Three "map" files are available in a compressed file (pennTB_tipster_wsj_map.tar.gz) as an additional download for users who have licensed Treebank-2 and provide the relation between the 2,499 PTB filenames and the corresponding WSJ DOCNO strings in TIPSTER....." 1574.6.9 Main advantage of Formal LLM is its deeplearning ability to synthesize sentences from text-knowledge graphs adhering to formal language theory with no machine learning statistical artifacts e.g large training corpus, Bayesian probability priors, billion sized neural network parameters among others. 1574.7 Following are some of the semi-filled sentences synthesized by earlier algorithm (unfilled PoS slots are marked with names in caps and duplicate tags are allowed) for query "what is electricity made of" - from testlogs/QuestionAnswering.log.7January2025: WordNet-ConceptNet TextGraph Random Walk - Bot generated answers ranked by wordnet perplexity: {0.0010078105316200553: 'NNP NNP CD protons physical MD VB DT electricity IN ' 'DT JJ NN NNP CD', 0.001149177619765855: 'NNP NNP CD electrons physical MD VB DT electricity IN ' 'DT JJ NN NNP CD', 0.001282667949334616: 'NNP NNP CD protons physical MD VB DT phenomenon IN DT ' 'JJ NN NNP CD', 0.0014109347442680777: 'NNP NNP CD protons physical MD VB DT stationary IN DT ' 'JJ NN NNP CD', 0.0014625896978838157: 'NNP NNP CD electrons physical MD VB DT phenomenon IN ' 'DT JJ NN NNP CD', 0.0016088486676721974: 'NNP NNP CD electrons physical MD VB DT stationary IN ' 'DT JJ NN NNP CD'} WordNet-ConceptNet TextGraph Random Walk - Bot generated answers ranked by wordnet perplexity: {0.14285714285714285: 'NNP NNP VBZ electricity IN NNP NNP DT NNP moving NN', 0.18181818181818182: 'NNP NNP VBZ phenomenon IN NNP NNP DT NNP moving NN', 0.2: 'NNP NNP VBZ stationary IN NNP NNP DT NNP moving NN'} |
|
More...
|
almost 2 years ago
|
| Section 1476 updated |
|
More...
|
almost 2 years ago
|
| Section 1459 updated |
|
More...
|
almost 2 years ago
|
| Section 1459 updated |
|
More...
|
almost 2 years ago
|
| Section 1459 updated |
|
More...
|
almost 2 years ago
|
| Section 1476 updated |
|
More...
|
almost 2 years ago
|
| +1573. (THEORY and FEATURE) Nachiketas Formal LLM - Dynamically configured Sentence PoS Dictionary retrieved by inorder traversal of grammar trees from treebank - Changes to Wikipedia Question-Answering algorithm for better accuracy 12 - 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 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 - 5 January 2025 |
|
More...
|
almost 2 years ago
|
