| [git reset to 28 April 2017 revision and recommitted] |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| ---------------------------------------------------------------------------------------------------------------------------------------- Commits - 28 April 2017 - NeuronRain AsFer-Boost::Python-C++ - VIRGO64 systemcalls and drivers invocations ---------------------------------------------------------------------------------------------------------------------------------------- *) VIRGO64 clone, kmemcache and filesystem system calls were invoked from Python-C++ boost extensions again and reproducibly work without kernel panics. *) Logs, persisted disk file written by filesystem system calls and rebuilt boost-python C++ extensions have been committed to testlogs/ |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| ---------------------------------------------------------------------------------------------------------------------------------- Commits - NeuronRain AsFer Boost C++-Python - VIRGO64 System calls + Drivers invocation - 26 April 2017 ---------------------------------------------------------------------------------------------------------------------------------- *) VIRGO64 system calls are invoked from C++ by Boost::Python extensions *) Kernel Logs for VIRGO KMemCache, Clone and FileSystem Calls Boost::python invocations have been committed to testlogs/ *) Boost version used is 1.64.0 *) setup.py has been updated with library_dirs config variable *) virgofstest.txt for filesystem calls has been committed to testlogs/ |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated asferpythonextensions.build.out for VIRGO64 |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| ------------------------------------------------------------------------------------------------------------------------------ Commit - NeuronRain AsFer-CPython Extensions VIRGO64 system call invocations - 20 April 2017 ------------------------------------------------------------------------------------------------------------------------------ *) VIRGO64 system calls are invoked from Python code by CPython extensions. *) Separate folder cpython_extensions64/ has been created for 64bit VIRGO kernel *) Requires Python 64-bit version *) Logs for VIRGO64 memory and filesystem systemcalls-to-drivers have been committed to testlogs *) With this Complete Application Layer-SystemCalls-Drivers request routing for python applications deployed on VIRGO64 works. |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| --------------------------------------------------------------------------------------------------------- Initial commits for Indexless Hyperball Recursive Web Crawler --------------------------------------------------------------------------------------------------------- *) An indexless crawler which tries to find a url matching a query string in a recursive crawl creating a hyperball of certain radius *) This is similar to Stanley Milgram, Kleinberg Lattice Small World Experiments and recent Hyperball algorithm for Facebook graph which concluded that Facebook has 3.74 degrees of freedom. *) Starts with a random url on world wide web. *) Success probability depends on average number of hops over large number of queries. *) This is a Randomized Monte Carlo Crawler and does a Depth First Search from an epicentre pivot url to start with *) Hyperball References: 1) http://webgraph.di.unimi.it/docs/it/unimi/dsi/webgraph/algo/HyperBall.html 2) Facebook Hyperball - https://arxiv.org/pdf/1308.2144.pdf |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| *) This commit implements an inverted index on top of ThoughtNet *) ThoughtNet is a hypergraph stored in file and on Neo4j Graph Database *) JSON loads ThoughtNet edges and hypergraph files and queries by classes returned by Recursive Gloss Overlap classifier *) Querying Neo4j is also better, but file representation of ThoughtNet is quite succinct and scalable because it just numerically encodes each edge (= a web crawled HTML document) *) file representation of ThoughtNet views the hypergraph as stack vertices interconnected by document id(s) similar to a source code versioning system. Having similar view on a graph database doesn't look straightforward. |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| ---------------------------------------------------------------------------------------------------------- Index implementation for NeuronRain AsFer Text and Crawled Web Documents Intrinsic Merit Ranking ---------------------------------------------------------------------------------------------------------- *) This commit adds a Locality Sensitive Hashing based inverted index for scrapy crawled HTML website pages. *) It is a derivative of LSH based similarity clustering implemented in python-src/LocalitySensitiveHashing.py *) It has to be mentioned here that ThoughtNet is also an inverted index for documents with additional features (i.e it is classifier based than raw string similarity) *) LSHIndex uses Redis Distributed Key Value Persistent Store as hashtable backend to store index. *) LSHIndex is not string to text inverted index, but hashed string to text inverted index (something like a digital fingerprint) *) Presently two types of hashes exist: primitive ordinal and MD5 hash *) Only one hash table is implemented on Redis with multiple hash functions with random choice. *) Presently 50 hash functions with 50 entries hash index has been implemented. |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
| Updated AsFer Design Document |
|
More...
|
over 9 years ago
|
