C2PO - ontology of acts related to content of clinical guidelines
Room 064: Need to setup server connection and thin clients (Hans is going to ask), Ellen's waiting for repsonse to give us access to the room
Jaspreet and Roar:
Going to add the sentence splitter (by Hans)
Have removed names, but the dates are still there
To finish their pre-prosessor and present it next week
Rune:
Working with Per in Tromsø, he's visiting next week; wednesday and thursday
Standoff Manager
Ingrid:
Looking at Thomas' algorithm for learning (training algorithm used for automatic classifications)
Discuss the training algorithm with Hans
Have tested the sentence splitter by Hans
Tags CVK-related events on clinical texts
To use Java instead of Python for post-tagging
Adverse event detection
Perry:
Ranks guidelines with information from EPJ (anonymized by Jaspreet and Roar)
Send information regarding DIPS access to Rune, then put it on the wiki
Everyone who has signed the NDA could access this system, if wanted
Hans:
Information retrieval
Sentence splitter
Summarization of the trajectory of a pasient
Cross-lingual interface between the EPJs and clinical guidelines
Sentence splitter:
Jaspreet and Roar are going to use a unmodified version Hans' sentence splitter
Hash map, checks characters and searches for dot (punktum) and looks for new line and abbreviations
Regex is the source in what Jaspreet and Roar does
Use the post-tagger for lemmatization as the next iteration? In need of lemma-lists (look for open alternatives)
Section based post-tagging before sentence splitting?
Output: Files with several sentences, with 0 as starting index
Separate standoff and xml tags
Could keep several standoff-files for different purposes
Common tools:
Name Entity Recognition
All should use BRAT or something similar to view the annotations from Jaspreet and Roar
Overlapping annotations are too complex for Notator and Semantator
Code to put tags back in text, so that Hans could use it in an algorithm