Differences

This shows you the differences between two versions of the page.

Link to this comparison view

deg:fall-2013-minutes [2015/06/12 16:46] (current)
ryancha created
Line 1: Line 1:
 +'''​Winter 2014 Meetings:'''​ Mondays at 3:​30-4:​30pm,​ Starting Jan. 6
 +
 +'''​Dec. 17, 2013:'''​
 +* Finals Week: those who can, come; those who can't, no pressure
 +* Joseph: made some minor updates to heuristic code, etc.; finished most of FROntIER rule editor
 +* Dr. Liddle: new granddaughter! Has to-dos with paper and annotator
 +* Thomas: coding and debugging (on step 9 of about 20 in the pipeline)
 +* Dr. Woodfield: working on agreement of definitions for "power levels"​ in conceptual modeling
 +* Dr. Embley: family history lunch; spoke with Dave Brown at tech transfer -- we'll license our technology to Church for $0 cost
 +'''​Sidebars:'''​ which pages are we going to process; ​
 +
 +'''​Dec. 10, 2013:'''​
 +* Dr. Liddle: progress on logging
 +* Peter: wrestled with LG parser, planned algorithm for semantic work, worked on apps for PhD programs
 +* Dr. Lonsdale: also worked on obituaries data; need to be able to publicly serve obituaries
 +* Tae-Woo: had been working on proposal, but found an issue to resolve first; fixed issue and back on proposal
 +* Thomas: continuing with simplified grammar induction approach -- reimplemented first few of several steps
 +* Shin: outlining paper for ER 2014; working on SQL queries over multiple databases; schedule: will leave USA on Jan. 15
 +* Dr. Embley: negotiate schedule for Winter 2014; Family History lunch; chapter officially accepted (sidebar)
 +'''​Sidebars:'''​ chapter accepted
 +
 +'''​Dec. 3, 2013:'''​
 +* Plan on meeting with Church History group at 2:30pm
 +
 +'''​Nov. 26, 2013:'''​
 +* Happy Thanksgiving! ​ No research meeting today.
 +
 +'''​Nov. 21, 2013:'''​
 +* Dr. Embley is doing the CS colloquium at 11:00am.
 +
 +'''​Nov. 19, 2013:'''​
 +* Tae-Woo: Working on proposal; almost done; hopes Dr. Embley feels the same. ;-)
 +* Joseph: Worked on improving rel-phrase heuristics, specifically for obituaries, to use actual matched text; needs further discussion with Dr. Embley perhaps; also worked on version numbers for FROntIER 1.1 and OntoES 2.4; HyKSS still works, but with OntoES 2.3
 +* Dr. Lonsdale: sent around newer listing of randomized pages; still needs to do skip pages piece
 +* Peter: code is working for a couple of simple examples to go all the way through, end-to-end; needs to discuss some elements with Dr. Lonsdale
 +* Shin: found that RDB annotator has some problems; there are some object- and relationship-sets that aren't being recognized; still working on SQL generator; needs to sketch out a paper; Dr. Embley has feedback for Shin from ER2013
 +* Thomas: might be done with the math he's been working on for several weeks; has finished several proofs; next step is to try to do the empirical part; still trying to follow up on Dr. Barrett'​s suggestion to meet with Dr. Ringger
 +* Dr. Liddle: ER 2013 was nice; working with Christopher on annotation tool
 +* Dr. Embley: ditto; found some issues with the annotator -- some form isn't working right; has written some things in the backlog; also found an issue with relationship-set display in the Ontology Editor
 +'''​Sidebars:'''​ sync up for annotation project
 +
 +'''​Nov. 5, 2013:'''​
 +* Peter: working on code; most is working; but progress feels slow (2 days jury duty, GRE, oral & written Mandarin midterms); needs to apply this month to PhD programs
 +* Tae-Woo: working on proposal
 +* Thomas: still working on minimum description length principle; trying to code what he's seen in other papers; can understand why what others have done won't work in his situation; is focusing on the mathematics behind it
 +* Shin: ran into some problems processing free-form queries (HyKSS not recognizing object or relationship sets in heterogeneous context); isn't quite sure how to make HyKSS do what he needs it to do; needs to identify whether the problem is in his code or in the HyKSS code
 +* Dr. Woodfield: working on extensions to conceptual models to handle family history concepts; e.g. crisp cardinality constraints in CM vs. distributions with confidence intervals in family history context
 +* Dr. Embley: continued working on keynote; continues to work on obituary project
 +* Dr. Lonsdale: has another set of random page ranges for us; is aware of every-other-page-blank situation now
 +* Dr. Liddle: getting ready for ER2013; Matthew Sheets did submit an ORCA proposal
 +'''​Sidebars/​blockers:'''​ Tae-Woo & Thomas (negative example -- false positive -- where did precision fail?); Shin & Dr. Embley at 3:30pm; ground-truthing tool
 +
 +'''​Oct. 29, 2013:'''​
 +* Plan to hear dry run of Dr. Embley'​s keynote presentation
 +
 +'''​Oct. 22, 2013:'''​
 +* Dr. Embley: spending almost all time on keynote presentation;​ up soon: training missionaries (four forms for missionaries are on Joseph'​s account)
 +* Joseph: looked at Sifter-produced docs; worked on thesis
 +* Peter: has been working on thesis code; would like to sit in on annotation training session
 +* Dr. Lonsdale: has a dozen students champing at the bit to annotate some things (could do first attempt at the task we're going to eventually hand the missionaries);​ working with a lot of other people; wants to discuss next stage for Sifter
 +* Shin: still working on generating SQL queries from free-form queries; needs to figure out how to combine multiple databases (union and join issues)
 +* Thomas: working on grammar induction (continues forever and ever!); has learned a fair amount from this week's experiments (memory issues); will work on making sure new grammar rule formula will work for future grammar rules
 +* Dr. Woodfield: appreciated discussion from last week with the group; has vocabulary question about his philosophical paper: is there a word that describes the fact that we can only talk about things that our language can express? ​ (Don't use Sapier-Whorf hypothesis, but this is in the neighorhood;​ perhaps the "frame problem"​ is somewhat in the neighborhood as well.)
 +* Dr. Liddle: resurrection demo; has been working with Christopher so the original Ontos (.rev2) runs again; now focusing on web-based annotation tool
 +
 +'''​Sidebars/​Blockers:'''​ Where are we with annotation, what's next?
 +
 +'''​Oct. 15, 2013:'''​
 +* Dr. Embley: finally got down to Moab to visit Arches with his family! ​ ML-OntoES paper revision is submitted
 +* Joseph: Look at JsonToOsmx, disabled unit tests to allow build to complete successfully,​ appears to be working aside from faulty unit tests; did some work on object existence rules for specialization objects
 +* Dr. Lonsdale: has two sample sets; worked on ML-OntoES paper; worked with Peter
 +* Dr. Liddle: worked on ML-OntoES paper; worked with Christopher (Java done, C++ infer program nearly done; next up: web programming);​ worked with Michael Bean
 +* Peter: has a new granddaughter! spent time refactoring SOAR bits; did some work on Big Data bits
 +* Tae-Woo: tried to tackle ontology snippet problem; worked on rewriting input to a simpler form
 +* Thomas: has been working on debugging and fixing grammar induction issues (micro issues, not macro)
 +* Shin: working on SQL generator; checking whether it works on a set of pre-formed queries across multiple databases with different schemas; looking for efficient new methods for generating keyword phrases over object sets in specific domains
 +* Dr. Woodfield: working further on extending conceptual modeling project; looking at intersection of family history and conceptual modeling
 +
 +* Discussion of "​right"​ label for modeling in a human way, not a programmer way; mention of Alan Newell'​s book (Unified Theories of Cognition) and the use of the term "​band"​ (Newell defines four bands: neural, cognitive, rational and social)
 +
 +'''​Sidebar:'''​ Steve & Deryle, Steve & Thomas
 +
 +'''​Oct. 8, 2013:'''​
 +* Dr. Embley: paper submitted, progress on OntoES-ML paper, interesting obituaries task to address (NDA)
 +* Joseph: made some progress on JSON-to-OSMX,​ but it needs more work (doesn'​t produce expected)
 +* Peter: finished thesis proposal, made more progress on writing; back to coding
 +* Dr. Lonsdale: working with Peter, retrained Sifter; needs input on what to train on and what pages to randomly pick (estimates 2/3 good books in the set of 200); has a dozen students in Linguistics Tools class ready to do annotation
 +* Shin: adopting new method for handling keyword phrases; has tried to compile LaTeX independently of his system, but no luck so far
 +* Tae-Woo: working on how to handle many patterns; it's a challenge to deal with many patterns; trying various approaches
 +* Thomas: continuing to work on redesigning pattern-finding code; getting closer, but it's slow [reminder: will need to set up meeting with committee in December]
 +* Dr. Woodfield: lots of interesting side issues are popping up (the pyramid isn't as clean as hoped)
 +* Dr. Liddle: reprocessed PDF examples; Founders Conference, E-Week were a success, progress with Christopher on old Ontos (not OntoES)
 +
 +'''​Sidebars:'''​ OntoES-ML paper
 +'''​Blockers:'''​ object-existence rules for specializations problem; font or character set in obituaries (Spanish characters) giving trouble in using Frontier (may need to run through iconv to convert to UTF-8)
 +
 +'''​Oct. 1, 2013:'''​
 +
 +* Presentation from Dr. Lonsdale on sifter process
 +* Shin: fixed some bugs, updated SQL generator so it generates UNION operations for gen/spec and SQL statements for INNER JOIN
 +* Dr. Woodfield: worked on philosophical view on Family History Research
 +* Dr. Embley: got Frontier running, ran in to some bugs/​blockers that Joseph knows about (or will know about soon); couldn'​t run directly from PDF; needs to try version from Jenkins; revived the product backlog
 +* Dr. Lonsdale: working with Peter
 +* Joseph: did some work on help parameter for Frontier; got sidetracked otherwise
 +* Dr. Liddle: hired Dr. Woodfield'​s student, made progress on getting him going in the porting project; discovered some issues with Antlr; etc.; worked with Michael Bean on other project
 +* Peter: finalizing thesis proposal; trying to get clarity on the mapper; wrote up ideas on evaluation component
 +* Tae Woo: working on issue of having 1000s of patterns; has some ideas to discuss
 +* Thomas: working through some data structures and algorithms on grammar induction; has almost given up on Kae build problems
 +
 +'''​Sidebars:'''​ almost out of time on the paper, Peter'​s thesis, Jenkins Kae build
 +
 +'''​Sep. 24, 2013:'''​
 +
 +* Dr. Woodfield: has a solution in search of a problem and some interesting directions to study
 +* Tae Woo: working on GreenFIE project, found some cases where it makes erroneous suggestions and other issues; made pattern to recognize paragraph info structure; that pattern could be useful for processing next page in Ely book
 +* Thomas: working on coding, pattern-finding,​ bug-fixing, etc.  Needs to figure out how to export png in graph tool.  Questions on Jenkins.
 +* Shin: new daughter!! Still recovering from long hospital stay and Mr. Mom duties. ​ (Congratulations,​ Shin!)
 +* Peter: working on clarifying algorithm for semantic/​linguistic data matching; hopes to send something to Dr. Embley before Thursday (1152 TMCB for Thu mtg)
 +* Dr. Embley: trying to run software; things didn't work out with annotator.ss;​ working with Joseph on trying to run Frontier; wants to make progress on obituary extraction project
 +* Joseph: worked with Dr. Embley on Frontier end-to-end run; close, but a few more changes needed
 +* Dr. Lonsdale: installed French Obituary folder and ran into troubles; now running; worked on sifter and produced linked output file of valid data (3-page runs, sometimes several in a good doc)
 +* Dr. Liddle: worked on Samantha'​s & Stephen'​s code for annotator; worked on old Gold Annotator; checked on old Ontos.rev2 code (4244 l.o.c.)
 +
 +'''​Sidebars/​Blockers:'''​ annotator web tool, obituary extraction process
 +
 +'''​Sep. 17, 2013:'''​
 +
 +* Dr. Liddle: worked on issues with PDFIndexer, obituaries, Jenkins
 +* Dr. Woodfield: worked on things not related to this group
 +* Dr. Embley: trying to get set up, code running, etc., thinking about obituary processing; worked on ML-OntoES paper
 +* Joseph: got sick last week, didn't get as much accomplished;​ uploaded workbench jar file into Google Drive
 +* Peter: working on OntoSOAR, installed some tools on his machine and Linguistics Lab machine (Eclipse plugin for SOAR); wrote some multi-sentence Java code
 +* Dr. Lonsdale: improved OntoSOAR parser; made progress on ML-OntoES paper; set up the classification system -- the sifter -- and found issues with PDFIndexer; ported sifter to supercomputer (mostly); how do we report 3-page ranges?
 +* Tae Woo: working on TWK -- possibly GreenFIE; working through some of the synergistic interaction questions
 +* Thomas: working on grammar induction still; ​
 +
 +'''​Sidebars:'''​ obituaries, sifter, Jenkins
 +'''​Blockers:'''​ Jenkins isn't cooperating (Thomas to look into this)
 +
 +'''​Sep. 10, 2013:'''​
 +
 +* Joseph: looked over David'​s code, found conflicts on update, is working on resolution; not as simple as just commenting out test cases; has more of that to do
 +* Dr. Liddle: has been working on ER 2013 travel arrangements,​ PDF indexer -- blocker with splitting PDFs
 +* Dr. Lonsdale: has made progress on his part of the paper (draft); needs to get together with Steve; may have till Oct. 13
 +* Peter: is looking at Harwood book, combination of parser & segmenter still has problems with the sample texts; will visit with Dr. Lonsdale; also has reconstructed summer code, getting some matches
 +* Thomas: working on grammar induction, debugging
 +* Dr. Woodfield: has continued discussions on paper -- talking about power, why extend/add to conceptual models; when does a construct become first class
 +* Dr. Embley: back in town; need candidates to hire as RAs
 +
 +'''​Sidebars:'''​ resolve PDF splitting questions, end-user interface
 +'''​Blockers:'''​ PDFIndexer, HyKSS demo, other demos?, Joseph needs data, Dr. E has trouble with WinSCP
 +
 +'''​Sep. 3, 2013:'''​
 +
 +* Peter: spent summer at SOAR Tech, gave brief report on experiences
 +* Dr. Lonsdale: ported family history classifier on dithers; ready to process text files (blocked by PDFIndexer multi-page tool); remember 9/21 deadline for paper; teaching undergrad class on linguistics tools -- would like to get students on the annotator tool
 +* Tae Woo: has been in Korea, working on personal and family business; back in the saddle now; has 2 classes this semester
 +* Thomas: presented HIP paper a couple weeks back (nice job!), Jake offered to help get more data; working on extending simple list grammar induction process to handle more complex record types
 +* Shin: almost done with SQL generation, but has some bugs to fix, adapting WordNet similarity for Java to handle keyword phrases
 +* Dr. Embley: attended HIP and ICDAR: someone is doing online handwriting recognition (could apply to detection of novel features, like early signs of Alzheimer'​s or autism -- often the things we do have far-reaching implications,​ beyond what you first think); wanted to tackle the problem of grand unification in terms of pattern recognition (Rejean Plamondon, Ecole Polytecnique de Montreal); Golden Age of Manuscript Studies and Imaging Technologies (reuse of papyrus, ability to pick up various writing layers with multi-spectral imaging); ICDAR had competitions -- we should see if we can sponsor one (similar to WEPS, like what we're doing with Church now); talked with ABBY rep there, and ability to zone -- should be able to handle it; HIP would like to be more practical in the sense of being useful to end users; needs HyKSS demo restarted
 +* Joseph: has committed changes to evaluator on rev. 1.1 so original evaluator; blocker: needs data
 +* Dr. Liddle: read through reviewer comments on HyKSS paper and read recommended additional references, needs to spend more time on PDFIndexer and resetting the demo tools
 +* Dr. Woodfield: is working on domain-specific conceptual modeling; (1) conceptual models don't let you represent certain things (e.g. probability distributions on participation constraints),​ (2) conceptual models are quite crisp, but we want to be able to override constraints sometimes, (3) evidence/​probability modeling constructs are not first-class features; primary question is "what are the problems with current conceptual models?"​
 +
 +'''​Sidebars:'''​ sifter, main line (Liddle needs to deal with the blockers here)
 +
 +'''​Blockers:'''​ PDFIndexer, HyKSS demo, other demos?, Joseph needs data
  
deg/fall-2013-minutes.txt ยท Last modified: 2015/06/12 16:46 by ryancha
Back to top
CC Attribution-Share Alike 4.0 International
chimeric.de = chi`s home Valid CSS Driven by DokuWiki do yourself a favour and use a real browser - get firefox!! Recent changes RSS feed Valid XHTML 1.0