Differences

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

Link to this comparison view

deg:winter-2015-minutes [2015/06/12 16:41] (current)
ryancha created
Line 1: Line 1:
 +'''​Apr. 22, 2015:'''​
 +* Meet at 2pm, location TBA
 +* Tae Woo: finished with finals, done with coursework!! ​ w00t!  Fixed some bugs with GreenFIE-HD:​ some offsets were 0-based instead of 1-based and vice versa. ​ Still more bugs to work on, but it's coming.
 +* Christopher:​ spoke with Justin (parents both Tawainese, is a Taiwan RM) vis a vis Jiapu project; Justin will finish 240 during Summer Term; finished finals; fixed gray background problem; strange problem with hover handler (stopped working for some reason), but it's fixed by changing it to mouse-enter/​mouse-leave events; fixed another strange issue with highlighting that he fixed
 +* Dr. Embley: convert-to-GedcomX program now works; it will need some refactoring,​ but not quite yet; has bounding-box success! ​ Currently working on cross-tool integration;​ has found an off-by-one error in the OSMX document coming from somewhere upstream in the tool chain. ​ There will be a host of other issues. ​ Blocker: resource URL is not being passed downstream.
 +* Dr. Liddle: wrote final, gave finals, graded finals (still reconciling,​ wrapping up grading); Jenkins is working properly
 +* Dr. Woodfield: has been working on the build; is beginning to use the wrappers; other program runs fine; needs a large data set to be able to generate soft-constraint distributions (Dr. Embley will check on whether we can get this data)
 +* '''​Sidebars:'''​
 +** Demo COMET updates
 +** Tool chain inconsistencies/​issues
 +** Woodfield'​s build
 +** Directory structure adjustments (Liddle/​Embley)
 +** Liddle/​Almquist tomorrow at 10am (communicating with ListReader)
 +
 +'''​Apr. 15, 2015:'''​
 +* Happy tax day, everyone!
 +* Tae Woo: was busy with finals, but worked on OSMX to JSON project a bit; tried to update Alpha and had a failure.
 +* Dr. Packer: looked at the test failures in the one file, and was able to resolve the test case that seemed most important; will check in before end of day; needs to debug Dr. E's installation of ListReader
 +* Christopher:​ given ListReader'​s input, should be able to populate the fields now; annotation is working; send seems to be okay, now needs to integrate with backend service; now gets text from offsets (so ListReader doesn'​t have to supply text); this only works on the PDF document type; reduced PDF image growth rate so it doesn'​t jump too much; having trouble diagnosing the gray background-on-hover issue; there is a cosmetic issue with ListReader training
 +* Dr. Embley: got thin line finished with end-to-end tool; working on thick line; getting ready for big presentation
 +* Dr. Liddle: did a ton more refactoring,​ especially on OntologyEditor,​ but all throughout the Alpha project; worked with Dr. Embley on a number of programming issues
 +* '''​Sidebars:'''​
 +** Alpha update with Tae Woo
 +** Find issue with gray backgrounds;​ see ListReader cosmetic training issue
 +** Offline: Liddle/​Almquist should meet on backend ListReader integration (Christopher will schedule separately with Dr. Liddle)
 +** Liddle/​Embley look at OntoSoar on Linux
 +
 +'''​Apr. 8, 2015:'''​
 +Starting April 28, DEG meetings will be Tuesdays at 2pm.
 +* Tae Woo: busy with class projects this week; learned about "​expected value",​ precision/​recall,​ and thinks perhaps this could help with his project
 +* Dr. Packer: building Alpha as we speak; has Netbeans 8; will look at test-case errors next time
 +* Christopher:​ lots of class projects this week; found a way to populate form, but getting empty string when sending it back
 +* Dr. Woodfield: spent a lot of time learning the Ontology Workbench; needs to learn more about Netbeans now
 +* Dr. Embley: thin thread works all the way through: extraction from page to GedcomX!
 +* Dr. Liddle: lots of time refactoring;​ lots of time with other DEG members
 +* Dr. Lonsdale: made progress with OntoSoar, and now has committed; got locations working and can associate them with different types of life events; has christening and other events in new ontology working; still left: multiple-word locations (e.g. "Salt Lake City") and 4-ary relationships;​ will be gone the next four weeks
 +* '''​Sidebars:'''​
 +** Liddle/Kim: Netbeans 8 and Java 8
 +** Woodfield/​Liddle:​ wrappers-consuming project setup
 +** Liddle/​Embley:​ keep Reema'​s stuff??
 +** Liddle/​Embley:​ access to file assets while running on dithers (server) instead of workstation
 +
 +'''​Apr. 1, 2015:'''​
 +* Dr. Packer: working with Dr. Embley on ListReader
 +* Christopher:​ added "​Execute ListReader"​ button to UI; working on train ListReader and populate form
 +* Dr. Woodfield: got Netbeans running with JDK 8; unable to get into Jenkins
 +* Dr. Embley: quarterly goal for Q1 was to get a thin thread through all steps in the ensemble; hasn't quite completed the very last piece: GedcomX; Q2 goal is a full thread with all the tools (OntoSoar, ListReader, COMET, GreenFIE-HD,​ all producing GedcomX output)
 +* Dr. Liddle: worked with Drs. Embley and Lonsdale quite a bit; did a ton of refactoring work
 +* Dr. Lonsdale: finally got Maven, Soar, Ontos, and Netbeans all playing nicely with each other [next week is last Wednesday he'll be here for a while -- getting back in town May 6]
 +* Tae Woo: made progress on programming GreenFIE-HD;​ calculating precision and recall (working on some bugs still); has some recovery issues right now
 +* '''​Sidebars:'''​
 +** ListReader test cases for Java 8 (Liddle, Packer) -- AStar code no longer is relevant -- likewise CRF; likely need to work on HMM code: see rev. 5865, 5866
 +** Dr. Woodfield needs access to Jenkins (Liddle, Woodfield) -- done
 +** The Alpha refactor & wrappers (everyone)
 +** RegEx highlighting for COMET (Embley, Almquist, Liddle)
 +** User Story 4 for OntoSoar (Embley, Lonsdale, Liddle)
 +
 +'''​Mar. 25, 2015:'''​
 +* Dr. Lonsdale: made some progress on Soar setup -- version number was a big issue; still not to the point where we can build at-will and be fully integrated with the rest of the Alpha project tree
 +* Tae Woo: hasn't made much progress this week, but finished OsmxToJson so it emits correct JSON now
 +* Dr. Packer: has reduced DEG hours, and is spending more time with Dr. Embley to get ListReader to run on his machines
 +* Dr. Woodfield: decided that modifying the XSD won't be helpful; syntax checking at the XML level isn't particularly helpful; working on Netbeans install (noted Wiki instructions are old)
 +* Christopher:​ made the red delete button smaller; multiple fields turning grey doesn'​t have an easy solution (might be a Bootstrap issue); for populating fields with ListReader annotations,​ has an idea of how to get the fields, but it needs more discussion
 +* Dr. Embley: mounting a steep learning curve, getting into wrappers; have agreement with Randy on GedcomX interface; highlighting demo is working well
 +* Dr. Liddle: worked with Embley & Lonsdale on various implementation issues and the ensemble layout
 +* '''​Sidebars:'''​
 +** KAE build
 +** Netbeans build issues with Dr. Woodfield
 +** Christopher:​ how to get fields from labels
 +** Ensemble directory structure
 +
 +'''​Mar. 18, 2015:'''​
 +* Dr. Liddle: made good progress on PDFIndexer issues (fixed bug, redid character bounding boxes to match PDFBox 2.0.0-SNAPSHOT)
 +* Dr. Packer: finished testing HMM with production mode scripts; worked on 3 pages of example; looking for smooth transition path
 +* Dr. Woodfield: looked at the object wrapper stuff, made progress understanding it; looked at long-term view of how to use the constraint checking tool; needs to figure out how to use the code
 +* Christopher:​ Got rid of hover issues and added "​annoying popup" on navigation away from page; also resized the delete button
 +* Dr. Embley: almost fully resolved on GEDCOMX with Randy; worked with Liddle to access pages; doing lots of programming
 +* Dr. Lonsdale: spent too much time trying to get SOAR built properly in both Eclipse and Netbeans to no avail (will meet with Liddle on Friday)
 +* Tae Woo: just about finished coding for the import (needs about an hour's worth of work still); evaluator should be done within about a week after that; need to begin integration with COMET (will meet with Embley/​Liddle after next week's meeting)
 +* '''​Sidebars:'''​
 +** PDFIndexer changes
 +** ListReader (potential complication)
 +** Ensemble directory structure, URLs
 +
 +'''​Mar. 11, 2015:'''​
 +* Tae Woo: has done un-nest operation, bidirectionally;​ need to re-nest and emit JSON
 +* Dr. Packer: working on text string passing between ListReader & COMET; adding text is a bit harder than previously thought
 +* Christopher:​ resource ID seems to be the best way forward -- see sidebar; some bug fixes; working on 288, 299
 +* Dr. Woodfield: dynamic constraint checking may be difficult, and it could be lower-hanging fruit to run after saving from annotator?
 +* Dr. Embley: did a demo for John from PDF through FROnTIER, to COMET, merger, inferencing for gender etc., Duke to de-duplicate,​ final output through Peter'​s code to produce GEDCOMX (which isn't even close to what we want). ​ Nearly finished mock-up of the GEDCOMX that we do want.  Figured out how to get bounding boxes into it (4-corner relative or pixel-based). ​ We could actually take our work and put it into FamilySearch.
 +* Dr. Liddle: helped various people with programming;​ progress studying web components and Polymer (interesting possibilities for us); discussed architecture of Fe6 management console thoroughly with Dr. Embley
 +* Dr. Lonsdale: figured out that it's not OSMXDocumentLoader causing grief, but something in the classpath settings somewhere
 +* '''​Sidebars:'''​
 +** JSON being passed between COMET and ListReader (both Dr. Packer and Christopher have discussion items)
 +*** How important is it to have text string in what's sent to ListReader? Answer: not very.  Let's punt on passing text back and forth.
 +*** COMET needs to know source ID or resource object, or some way to get to it; using only a path will cause half of the other things to break that depend on the full resource.
 +** Hosting FamilySearch data extensions
 +** Current priorities
 +** Bounding boxes
 +** Refactoring projects in the SVN repository (e.g. OntoSoar bound up with other modules)
 +
 +
 +'''​Mar. 4, 2015:'''​
 +* Welcome back Peter! ​ And Liddle!
 +* Dr. Liddle: got some work done getting object-existence rules into the converter.
 +* Peter: happy about having thesis paper accepted to AAAI where it was well received.
 +* Dr. Lonsdale: coding frenzy - ocean = OntoSoar running on Eclipse Lunar; upgraded to Soar 9.4; got everything to build in Eclipse, also on NetBeans. ​ Had to comment out an OSMX test in loader code.  Document loader in OntoSoar has some issues with changes in the library. ​ Needs permission to commit code to SVN repository.
 +* Tae Woo: busy with class projects and being sick; working on unnesting records; next up will be sorting by offset, then dealing with parent1/​parent2 merge situations
 +* Dr. Packer: turned a version of ListReader to Christopher to play with; next up is adding text to the JSON
 +* Christopher:​ hasn't yet been able to use the new ListReader; just changed it to work with the new format; discussed using parameter instead of file with Dr. Packer; has ListReader highlighting working
 +* Dr. Embley: did some coding (not quite a coding frenzy, snow falling) dealing with the full line from PDF to GEDCOMX; will need to deal with jobs soon (directory structure, etc.)
 +* Dr. Woodfield: working on constraints,​ regular expressions,​ adding semantics
 +* '''​Sidebars:'''​
 +** SVN repo access for lonz
 +** Annotator2 popup on hover needs to go away
 +** How to deal with bounding boxes (begin by setting up a tool that post-processes OSMX files to read bounding-box info from XML files)
 +** What Dr. Embley is promising to hand over to Randy Wilson
 +** ListReader: demo, also Dr. Packer & Christopher on parameter discussion
 +
 +'''​Feb. 25, 2015:'''​
 +* Discussed Fe6 Architecture (quarterly goals and the current status of each in the progress reports below)
 +* Dr. Liddle: in Hawaii (but coded object-existence rules)
 +* Dr. Lonsdale: OntoSoar running under Eclipse, needs new target ontology
 +* Tae Woo: Coding osmx2json -- import done; unnest, merge, nest, export lacking
 +* Dr. Packer: got passed major bug, coordinating with Christopher about transferring info back and forth
 +* Christopher:​ highlighting largely working, will turn attention to ListReader interface
 +* Dr. Woodfield: questions about mismatch in conceptual models
 +* Dr. Embley: Fe6 architecture;​ object-existence rules (w/ Dr. Liddle), functional for single-entry fields
 +* '''​Sidebars:'''​
 +** (email) Christopher and Dr. Packer: ListReader interface via file transfer vs. parameter passing
 +** (next week) bounding boxes -- pass through vs. determine during FamilyTree generation
 +** (next week) artifacts during highlighting (two fields with identical annotations,​ focus field highlighted after initial hover, multiple entry fields grey after initial hover, pencil-+-x still shows in annotator2)
 +
 +'''​Feb. 19, 2015:'''​
 +* Dr. Liddle is out of town
 +* Discussed Luther Tychonievitch'​s RootsTech presentation:​ "From Purple Prose to Machine-Checkable Proofs: Levels of rigor in family history tools"
 +
 +'''​Feb. 12, 2015:'''​
 +* Today is RootsTech developer conference; no Dr. Embley today
 +* Dr. Liddle: IRB has approved my proposal; got some programming done with Dr. Embley; fixed OntologyEditor w.r.t. relationship connections;​ spent some time working with Christopher
 +* Dr. Embley presented at FHTW
 +* Tae Woo: has been programming on getting form JSON ready for the next phase in the pipeline; is implementing it in a way that works just for his thesis; issue: sometimes COMET freezes when advancing to next page; next up is getting various JSON structures finalized
 +* Christopher:​ hover highlighting is mostly working, but there are some special cases that still need work; deprecated standard mode, so forms now default to grid mode; has questions about adding to nested fields (two similar user stories may need coalescing)
 +* Dr. Woodfield: made progress on CS 340; nearing the end of screens, JS, backend; persistence is up next; two things from conference yesterday: (1) thinking about a mechanism to apply probabilistic constraints to the Church'​s One Tree to Rule Them All; (2) need a declarative mechanism for specifying rules and constraints. ​ It's time for a new modeling standard.
 +* '''​Sidebars:'''​
 +** Christopher:​ demo highlighting,​ user stories on nested fields
 +** Tae Woo: how to really convert to form JSON
 +
 +'''​Feb. 5, 2015:'''​
 +* Dr. Lonsdale: AAAI conference was really good (1000 people); our talk went well -- Peter did a great job
 +* Dr. Woodfield: still being inundated by CS 340; Dr. L & Dr. L may be able to serve as advisers to CS MS students
 +* Dr. Packer: has gotten back into the code, and BigInteger fix seems to be working and working reasonably quickly; will continue with change to JSON instead of OSMX, keeping track of labels to send info back to annotator to highlight different parts of query string
 +* Christopher:​ has highlight functionality working (calling function from console); finding what is needed in the DOM on hover is interesting,​ and needs some more work
 +* Dr. Embley: RootsTech development summit is next Wednesday, so won't be to DEG meeting; our FHTW workshop paper was accepted; working on a paper with George Nagy; did some programming;​ have object existence rules been deployed to annotator2? (needs to be checked)
 +* Dr. Liddle: worked with Dr. Embley on some bugs & coding issues; got IRB application submitted
 +* '''​Sidebars:'''​
 +** Work out example for Christopher to test highlighting
 +** Work out example for calling ListReader ???
 +** Embley presentation outline
 +* '''​Notes:'''​
 +** Liddle perhaps should fix OntologyEditor order bug for objects in relationship connections
 +** Liddle definitely should fix PDFIndexer/​annotator vertical axis issue
 +** Tool names:
 +*** The annotator will be called COMET: Click-Only (or at least Mostly) Extraction Tool
 +*** The form-based extraction tool ensemble will be called Fe5: Fe == Iron => strong; 5 is version number, and we have ideas for versions 6 and 7; here are the phases:
 +**** 1. Extract with multiple extraction tools
 +**** 2. Merge extracted records into filled-in forms
 +**** 3. Resolve: human looks at form, corrects as needed
 +**** 4. Accuracy check: we may just let the ensemble do the job automatically,​ or we may need to keep people involved
 +**** 5. Learning: feedback loop to improve the tools, and with Tae Woo's work learn new extraction rules so the tools keeps getting better
 +**** 6. Family Tree creation and check (Dr. Woodfield'​s work)
 +**** 7. Generate the full WoK-HD that we've talked about in the past (add back in the query components, etc.)
 +
 +'''​Jan. 28, 2015:'''​
 +* Dr. Lonsdale is in Texas with OntoSOAR paper
 +* Dr. Liddle: NSF proposal has advanced to "​recommended"​ status; working on IRB clearance now
 +* Tae Woo: has been programming;​ successfully converts OSMX to record, and now needs to convert records to JSON
 +* Dr. Packer: has a couple of papers ready to resubmit; next week will be back to programming ListReader
 +* Dr. Woodfield: same as last week (making progress)
 +* Christopher:​ has been working on hover highlighting;​ has something that should be highlighting,​ but just getting border, not color; general highlight is mostly working, and pretty close for single entries on highlighting;​ but hover isn't working
 +* Dr. Embley: has been working with Thomas on papers; got some programming and inference work done this week; will work on export to GedComX
 +* '''​Sidebars:'''​
 +** XSD import/​include (the difference between the two is a namespace issue)
 +** Highlight demo/​discussion,​ product backlog discussion
 +** Thinking of names for ensemble, annotator tool
 +** Thomas had some ideas on making the interface work a lot faster
 +
 +'''​Jan. 21, 2015:'''​
 +* Congratulations,​ Joseph, on passing your thesis defense
 +* Joseph: passed defense, got paperwork done
 +* Dr. Lonsdale: wants to sidebar on Maven problem (784 TNRB, 9:30am Thursday)
 +* Tae Woo: has been programming on OsmxToJson, especially converting record part; figured out some solutions, feels we have it; needs deploy of annotator2
 +* Dr. Packer: has been working with Dr. Embley on the paper; should be done next week with those; ListReader development follows
 +* Christopher:​ has split the highlighting story into three; has something for general highlighting and knows the basic idea for single entry and multiple entry
 +* Dr. Liddle: has been doing miscellaneous things and refactoring osmx2.xsd
 +* Dr. Embley: we have the Springer book Towards the Multilingual Semantic Web, submitted FHTW paper, met with Dr. Lonsdale; wants to talk about recruiting students
 +* Dr. Woodfield: still working on CS 340 for another few weeks
 +* '''​Sidebars:'''​
 +** Rejuvenating the group: could Scott be the pseudo adviser plus a real adviser
 +** OSMX2 .xsd discussion
 +** New flow for downstream that takes as input all the OSMX documents that are produced along the way (solves the problem with gender we had discussed)
 +
 +'''​Jan. 14, 2015:'''​
 +* Joseph to present thesis defense dry run
 +* Scrum meeting afterwards if time
 +
 +'''​Jan. 7, 2015:'''​
 +* Joseph: worked on presentation for defense enabled probability output features (incl. from command line)
 +* Dr. Liddle: got IS 411 started again; refined the deployment scripts; heads-up on maven assembly plugin and build process
 +* Dr. Lonsdale: got back yesterday
 +* Tae Woo: still struggling with converting OSMX to JSON
 +* Dr. Packer: working with Dr. Embley, responding to paper feedback, some more coding
 +* Christopher:​ getting back to classes; fixed adding new record with A/a key; working on highlighting for GreenFIE
 +* Dr. Woodfield: mostly has worked on CS 340
 +* Dr. Embley: working with Thomas on the paper; will resubmit soon
 +* '''​Sidebars:'''​
 +** Tae Woo's issue with OSMX-to-JSON conversion and page issues/page boundaries/​complex annotations/​edited text, etc.
 +*** Note: complex annotations need to be refactored in the OSMX xsd; JSON correctly associates resources with annotation components; OSMX does not.
 +*** Note: Dr. Embley needs to add a user story to fix JSON to OSMX to deal with AnnotatedComponent required attributes (id, type, inResource).
 +** Bug for appending annotation from multiple pages [NOTE: unable to replicate this issue on Mac/Chrome with annotator2]
 +** Think about FHT Workshop & Jiapu (can Dave write abstract?)
  
deg/winter-2015-minutes.txt ยท Last modified: 2015/06/12 16:41 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