161sh/CatgirlIntelligenceAgency
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
1,587 Commits 19 Branches 14 Tags 14 MiB
a0f28a7f9b
Commit Graph

60 Commits

Author SHA1 Message Date
Viktor Lofgren
e46e174b59 (keyword-extractor) Add another test for Name-extractor 2024-01-01 15:21:51 +01:00
Viktor Lofgren
1cbf23e7e7 (test) Don't fail test if atags.parquet is not in ~vlofgren 2023-11-15 09:11:38 +01:00
Viktor Lofgren
e0c769fd19 (converter) Integrate atags.parquet with the encyclopedia sideloader 
Also clean up stackexchange and dirtree a bit.
 2023-11-06 18:03:01 +01:00
Viktor Lofgren
ebd10a5f28 (crawler) Integrate atags.parquet with the crawler so that "important" URLs are prioritized 2023-11-06 16:14:58 +01:00
Viktor Lofgren
2b77184281 (converter) Integrate atags with the topology field 2023-11-06 13:46:44 +01:00
Viktor Lofgren
72afa0341f duckdb connection may need to be synchronized? 2023-11-04 14:30:25 +01:00
Viktor Lofgren
0152004c42 Initial Commit Anchor Tags 
* Added new (optional) model file in $WMSA_HOME/data/atags.parquet
* Converter gets a component for creating a projection of its domains onto the full atags parquet file
* New WordFlag ExternalLink
* These terms are also for now flagged as title words
* Fixed a bug where Title words aliased with UrlDomain words
* Fixed a bug in the encyclopedia sideloader that gave everything too high topology ranking
 2023-11-04 14:24:17 +01:00
Viktor Lofgren
4415f52e18 (keyword-extraction) Fix broken test 2023-10-27 12:19:33 +02:00
Viktor Lofgren
c51159672e (build) Move unit test configuration to root build.gradle 2023-10-04 12:46:22 +02:00
Viktor Lofgren
93dc80000c (bugfix) Fix NPE in KeywordExtractor due to bad SoftReference handling 2023-09-26 17:16:41 +02:00
Viktor Lofgren
9b781f8404 (keyoword-extractor) Address very rare race condition in memoization logic 2023-09-25 18:28:04 +02:00
Viktor Lofgren
8ca20f184d (keyword-extraction) Chasing my tail looking for a bug 2023-09-24 19:39:48 +02:00
Viktor Lofgren
dbe9235f3a (*) Upgrade to JDK21 with preview enabled. 
... also move some common configuration into the root build.gradle-file.

Support for JDK21 in lombok is a bit sketchy at the moment, but it seems to work.  This upgrade is kind of important as the new index construction really benefits from Arena based lifecycle control over off-heap memory.
 2023-09-24 10:38:59 +02:00
Viktor Lofgren
70aa04c047 (converter, stackexchange-xml) Add the ability to sideload stackexchange data 2023-09-21 12:48:33 +02:00
Viktor Lofgren
5b0a6d7ec1 (stackexchange-converter) Create tool for converting stackexchange 7z-files to digestible sqlite db:s 2023-09-20 15:15:13 +02:00
Viktor Lofgren
3b4d08f52b (stackexchange-integration) Add better comments 2023-09-20 14:43:06 +02:00
Viktor Lofgren
6bbf40d7d2 (stackexchange-integration) Tools for reading stackexchange xml files 2023-09-20 14:17:33 +02:00
Viktor Lofgren
eaeb23d41e (refactor) Remove converting-model package completely 2023-09-14 11:21:44 +02:00
Viktor Lofgren
c68d17d482 (keyword-extraction) Fix bug leading to position data missing on some keywords. 
This was due to a discrepancy between the KeywordPositionBitmask and WordsTfIdfCounts' concept of a keyword.
 2023-09-02 14:48:55 +02:00
Viktor Lofgren
676e7c7947 (keywords) Add Serializable properties that went missing as the record became a class 2023-09-02 09:52:01 +02:00
Viktor Lofgren
5f427d2b4c (keywords) Clean up leaky abstractions, clean up tests 2023-09-01 13:52:00 +02:00
Viktor Lofgren
3101b74580 (index) Move to a lexicon-free index design 
This is a system-wide change.  The index used to have a lexicon, mapping words to wordIds using a large in-memory hash table.   This made index-construction easier, but it
also added a fairly significant RAM penalty to both the index service and the loader.

The new design moves to 64 bit word identifiers calculated using the murmur hash of the keyword, and an index construction based on merging smaller indices.

It also became necessary half-way through to upgrade guice as its error reporting wasn't *quite* compatible with JDK20.
 2023-08-28 14:02:23 +02:00
Viktor Lofgren
ebc84c22fb Upgrade antique lombok plugin 
This permits tests to run on JDK20 environments.
 2023-08-23 14:34:32 +00:00
Viktor Lofgren
aa0d256d6a Upgrade code to Java 20. 
* Change language version
* Upgrade Lombok to a JDK20 compatible version
 2023-08-23 13:37:49 +00:00
Viktor Lofgren
db0216936e (summary) Reduce the chance of expensive operations 2023-08-16 15:48:34 +02:00
Viktor Lofgren
46d761f34f (language) fasttext based language filter 2023-08-16 15:48:12 +02:00
Viktor Lofgren
4ab1cd9502 (*) last touches 2023-08-07 12:57:44 +02:00
Viktor Lofgren
77d5e39fe0 Make processed data Serializable 2023-07-28 18:11:19 +02:00
Viktor Lofgren
baff83912e Small optimizations that shave an hour of processing time :D 2023-06-28 15:41:10 +02:00
Viktor Lofgren
f8f9f04158 Specialized logic for processing Lemmy-based websites. 2023-06-27 10:57:54 +02:00
Viktor Lofgren
7326ba74fe Tweaks to pub date heuristics to make it mostly get the 'historyofphilosophy.net' case right. 
Use HTML standard for plausibility checks in the more guesswork-like heuristics. Added more class names to look for date strings.
 2023-06-20 14:15:05 +02:00
Viktor Lofgren
32a6735d03 Undo change in requirements for counting as a high tf-idf word 2023-06-19 17:58:19 +02:00
Viktor Lofgren
f0b4acb358 Better logic for summarization. 2023-06-19 17:58:19 +02:00
Viktor Lofgren
9579cdd151 Improved heuristic for which words are considered important in selecting the summary text. 2023-06-19 17:58:19 +02:00
Viktor Lofgren
443cf0cf1e Expose additional functionality through WordsTfIdfCounts. 
Bump requirements for being flagged as high TF-IDF from 2 occurences to 3.
 2023-06-19 17:58:19 +02:00
Viktor Lofgren
4138233ddf Truncate repeated strings of any non-alnum symbols in SummaryExtractor 2023-06-19 17:58:19 +02:00
Viktor Lofgren
2979f4703e Allocation-free text utility 2023-06-19 17:58:19 +02:00
Viktor Lofgren
379bccc1a3 Disable AdblockSimulator since it's slow and doesn't really work. Just wasting CPU cycles until it's fixed. 2023-06-19 17:58:19 +02:00
Viktor Lofgren
21125206b4 Fix some bugs in JSON+LD-heuristics for pub date. 2023-06-19 17:58:19 +02:00
Viktor Lofgren
88399e30e2 Consider keyword relevance signals when creating the document summary using the DOM walker. 2023-06-19 17:58:19 +02:00
Viktor Lofgren
d82a858491 Don't consider slash to be a sentence separator. 2023-05-31 16:54:30 +02:00
Viktor
7694a15f62
Fix kale's unreasonably high weighting factor 2023-04-22 20:55:09 +02:00
Viktor Lofgren
619fb8ba80 (converter) Adjust the pub-date sniffing heuristics' order. Doing HTML5 tags too early puts some sites too early. Also expanded support for JSON+LD. 2023-04-19 15:28:50 +02:00
Viktor Lofgren
810515c08d Clean up artifact extractor. 2023-04-10 13:07:54 +02:00
Viktor
a278fc6296
Increase search result relevance (#8) 
* Increase accuracy of the position bits.
* Increase their width to 56.
* Use a rolling position scheme for bits 16-56 to increase the average accuracy.
* Result ranking overhaul
* Optimized queries
* BM25 in the index service's ranking
* Make gui less jank
* Javadocs for ranking parameters.
 2023-04-07 20:18:08 +02:00
Viktor Lofgren
716ab35b4e Search ranking debuggability improvements. 2023-04-02 13:43:24 +02:00
Viktor Lofgren
137adb9c3c Bitmask calculation improvement. Take sentence length into consideration, not all lines are equal. 2023-03-30 15:42:06 +02:00
Viktor Lofgren
0fcb2b534c Polish Names 2023-03-29 16:51:47 +02:00
Viktor
0b505939ed
Update features-convert/readme.md 2023-03-25 12:43:58 +01:00
Viktor Lofgren
46f81aca2f Break apart reverse index into a separate full index and priority index. It did this before using the same code. This will make the priority index about half as big since it no longer needs to keep metadata. 2023-03-21 16:12:31 +01:00