karlhigley lexrank-summarizer issues

karlhigley / lexrank-summarizer

A Spark-based LexRank extractive summarizer for text documents

MIT License

19 stars 4 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Use accumulators to quantify boilerplate removal

#44 karlhigley opened 8 years ago
0
Use GraphX .reverse method to generate bidirectional edges

#43 karlhigley opened 8 years ago
0
Use Spark's Dataframes API

#42 karlhigley opened 8 years ago
1
Maintain the order of excerpted sentences

#41 karlhigley opened 8 years ago
0
Add a link to the LSH paper that explains the pooling trick

#40 karlhigley closed 8 years ago
0
Update the README to reflect dynamic stopword filtering

#39 karlhigley closed 8 years ago
0
Remove obsolete option for number of LSH buckets

#38 karlhigley closed 8 years ago
0
Switch from a stopword list to dynamically identified stopwords

#37 karlhigley closed 8 years ago
0
Replace explicit removal of zeros with conversion to SparseVector

#36 karlhigley closed 8 years ago
0
Remove extraneous launch script

#35 karlhigley closed 8 years ago
0
Stop combining documents with the same identifier during pre-processing

#34 karlhigley closed 8 years ago
0
Revise description of SRP-LSH boilerplate filtering

#33 karlhigley closed 8 years ago
0
Add LSH cosine estimation method of graph building to LexRank

#32 karlhigley closed 8 years ago
0
Compute stopwords from the corpus on the fly

#31 karlhigley closed 8 years ago
0
Represent similarities as Floats instead of Doubles in LexRank

#30 karlhigley closed 8 years ago
0
Update to Spark 1.5.0

#29 karlhigley closed 8 years ago
0
Consolidate input document content by doc ID

#28 karlhigley closed 8 years ago
0
Rename CosineLSH to SignRandomProjectionLSH

#27 karlhigley closed 9 years ago
0
Update to Spark 1.4.1

#26 karlhigley closed 9 years ago
0
Add a configuration option for the number of LSH buckets

#25 karlhigley closed 9 years ago
0
Cache (LSH signature, feature vector) pairs

#24 karlhigley closed 9 years ago
0
Disentangle/test similarity computation and Lexrank model

#23 karlhigley closed 9 years ago
0
Test and refactor featurization code

#22 karlhigley closed 9 years ago
0
Add basic tests for CosineLSH model

#21 karlhigley closed 9 years ago
0
Precompute size of sparsified matrices (instead of auto-computation)

#20 karlhigley closed 9 years ago
0
Improve tokenization to reduce dimensionality

#19 karlhigley closed 9 years ago
0
Fix broken variable reference from repartitioning code

#18 karlhigley closed 9 years ago
0
Use minPartitions argument when reading file instead of repartitioning

#17 karlhigley closed 9 years ago
0
Revert "Combine input entries with the same identifier into a single …

#16 karlhigley closed 9 years ago
0
Properly parse input lines with text containing tabs

#15 karlhigley closed 9 years ago
0
Combine input entries with the same identifier into a single document

#14 karlhigley closed 9 years ago
0
Apply Kryo serialization

#13 karlhigley closed 9 years ago
0
Adjust spacing and driver class name in README

#12 karlhigley closed 9 years ago
0
Remove extraneous imports from LexRank model

#11 karlhigley closed 9 years ago
0
Uses locality-sensitive hashing to group similar sentences into buckets

#10 karlhigley closed 9 years ago
0
Move featurization to separate class and excerpt selection to Driver

#9 karlhigley closed 9 years ago
0
Amend script to allow 2g memory for driver and executors

#8 karlhigley closed 9 years ago
0
Straighten out package name, split into separate files

#7 karlhigley closed 9 years ago
0
Add corpus-level boilerplate filtering

#6 karlhigley closed 9 years ago
0
Add a link to the LexRank paper in the README

#5 karlhigley closed 9 years ago
0
Avoid creating graph edges between different documents

#4 karlhigley closed 9 years ago
0
Extract featurization into a companion object utility function

#3 karlhigley closed 9 years ago
0
Fix a typo in the previous edge filtering refactor

#2 karlhigley closed 9 years ago
0
Pre-filter the graph edges rather than filtering the graph itself

#1 karlhigley closed 9 years ago
0