July 17, 2009 NEMO Year 1: Overview & Planning .

49
July 17, 2009 NEMO Year 1: Overview & Planning http://nemo.nic.uoregon.edu

Transcript of July 17, 2009 NEMO Year 1: Overview & Planning .

July 17, 2009

NEMO Year 1:Overview & Planning

http://nemo.nic.uoregon.edu

Overview Agenda

• Introductions & go-to people (7 mins)• Scheduling regular teleconferences (3 mins)• Review of project aims (15 mins)• Contributing to NEMO -- overview (10 mins)

(website, wiki, database)• Overview of current ontologies (25 mins)• Overview of tools for labeling data (next time)

Action items highlighted in lime green!

Overview Agenda

• Introductions & go-to people (7 mins)• Scheduling regular teleconferences (3 mins)• Review of project aims (15 mins)• Contributing to NEMO -- overview (10 mins)

(website, wiki, database)• Overview of current ontologies (25 mins)• Overview of tools for labeling data (next time)

Action items highlighted in lime green!

Introductions: Who we are (1/3)

• NEMO “Core” (PIs & go-to people)– Dejing Dou (lead PI, CIS)– Gwen Frishkoff (co-PI, Psychology)– Allen Malony (co-I, CIS)– Don Tucker (co-I, Psychology)– Paea LePendu* (Ontology Development)– Robert Frank* (EEG/ERP Analysis Tools)– Jason Sydes* (Database & Wed Portal)– Haishan Liu (Grad Student, CIS)

• Matt Cranor & Charlotte Wise (Grants Admin)

Introductions: Who we are (2/3)

• NEMO Consortium– John Connolly (McMaster U)– Tim Curran (U Colorado)– Joe Dien (U Maryland)– Kerry Kilborn (Glasgow U)– Dennis Molfese (U Louisville)– Chuck Perfetti (U Pittsburgh)

• Please send link to your website to Jason ([email protected])

Introductions: Who we are (3/3)

• External collaborators (NEMO ontologies & database development; integration with other projects in BO community)– Jessica Turner (fBIRN & “CogPO” project)– Angela Laird (BrainMap & “CogPO” project)– Maryann Martone (NIF -- www.neuinfo.org)– Jeff Grethe & Scott Makeig (“HeadIT” project)– Folks at OBOF (http://www.obofoundry.org/)?– Folks at NCBO (http://bioontology.org/)?

Overview Agenda

• Introductions & go-to people (7 mins)• Scheduling regular teleconferences (3 mins)• Review of project aims (15 mins)• Contributing to NEMO -- overview (10 mins)

(website, wiki, database)• Overview of current ontologies (25 mins)• Overview of tools for labeling data (next time)

Regular Meetings

• Schedule using Doodlehttp://www.doodle.com/

• Once monthly?• Gwen to propose dates & times on Doodle for

next month’s meeting later today• Please respond to Doodle email (click on link

and check available days & times)

Overview Agenda

• Introductions & go-to people (7 mins)• Scheduling regular teleconferences (3 mins)• Review of project aims (15 mins)• Contributing to NEMO -- overview (10 mins)

(website, wiki, database)• Overview of current ontologies (25 mins)• Overview of tools for labeling data (next time)

Overview of Project Aims1. Design and test procedures for automated ERP pattern

analysis and classification (*)– “top-down” initial definitions of pattern rules, concepts

(hypotheses)– “bottom-up” data mining for pattern validation & refinement

2. Capture rules, concepts in a formal ERP ontology (TODAY)3. Develop ontology-based tools for ERP data markup (*)4. Apply ERP analysis tools to consortium datasets (*)5. Perform meta-analyses of consortium data (*)6. Build relational database to store ontology-based

annotations and to support complex reasoning over annotated data

“ontology database”7. Build data storage & management system

“EEG database”

(*) Proposed focus of next month’s meeting

The three pillars of NEMO

• Ontologies (TODAY)• Ontology-based analysis tools (next time?)• Ontology database & portal

Overview Agenda

• Introductions & go-to people (7 mins)• Scheduling regular teleconferences (3 mins)• Review of project aims (15 mins)• Contributing to NEMO -- overview (10 mins)

(website, wiki, database)• Overview of current ontologies (25 mins)• Overview of tools for labeling data (next time)

NEMO Centralnemo.nic.uoregon.edu

Contributing to NEMO• NEMO central

– http://nemo.nic.uoregon.edu• NEMO ftp site (EEG database)

– ftp://nemo.nic.uoregon.edu/EEG_Experiments• NEMO sourceforge (ontologies)

– http://nemoontologies.svn.sourceforge.net/viewvc/nemoontologies/current/

• NEMO listserve (to note ontology “bugs” and feature requests)– http://sourceforge.net/mail/?group_id=263320

• NEMO wiki (discussion) – coming soon…

Overview Agenda

• Introductions & go-to people (7 mins)• Scheduling regular teleconferences (3 mins)• Review of project aims (15 mins)• Contributing to NEMO -- overview (10 mins)

(website, wiki, database)• Overview of current ontologies (25 mins)• Overview of tools for labeling data (next time)

Why (what problem are we trying to solve?)

What (what IS an ontology anyway, and how can it help address this problem?)

How (ERP ontology design and implementation methods in NEMO)

NEMO ontology development

Why are there so few

statistical meta-analyses

in ERP research?

The Problem

410 ms

450 ms

330 ms

Peak latency 410 ms

Loose Semantics!

Will the “real” N400 please step forward?

Sample Database Query: Show me all the N400 patterns in the database.

Putative “N400”-labeled patterns

Parietal N400

≠≠

Frontal N400

Parietal P600

What’s an ontology and how does it help us address the lack of integration in ERP

research?

Ontologies to support VALID pooling of ERP patterns across

datasets theoretical integration

Why ontologies in particular?

Rich, explicit, computable semantics…. But takes time to build!

How we’re going to build ontologies for NEMO

[…and apply them to real data – next time]

FIRST RELEASE OF ONTOLOGIES IN AUGUST (DON’T BOTHER TO

COMMENT ON OLD VERSIONS…)

NEMO ontology design principles(following OBO “best practices”)

1. Factor the domain to generate modular (“orthogonal”) ontologies that can be reused, integrated for other projects

2. Reuse existing ontologies (esp. foundational concepts) to define basic (upper & mid-level) concepts

3. Validate definitions of complex concepts using bottom-up (data-driven) as well as top-down (knowledge-driven) methods

4. Collaborate with a community of experts in collaborative design, testing of ontologies

Factoring the ERP domain

1 sec

TIME SPACE

FUNCTION Modulation of pattern features (time,

space, amplitude) under different experiment conditions

ERP spatial subdomain

1 sec

TIME SPACE

FUNCTION Modulation of ERP pattern features under different experiment conditions

International 10-10 EEG Electrode Locations

ITT electrode location Fz

(medial frontal)

Scalp surface “regions of interest”

NEMO Spatial Ontology

BFO (Basic Formal

Ontology) UPPER

FMA(Foundational

Model of Anatomy ontology)

MIDLEVEL

SNAP

ERP temporal subdomain

1 sec

TIME SPACE

FUNCTION Modulation of ERP pattern features under different experiment conditions

Early (“exogenous”) vs. Late (“endogenous”) ERP processes

~0-150 ms after event (e.g., stimulus onset)

501 ms or more after event (e.g., stimulus onset)

~151-500 after event (e.g., stimulus onset)

EARLY

LATE

MID-LATENCY

NEMO Temporal OntologySPAN

ERP functional subdomain

1 sec

TIME SPACE

FUNCTION Modulation of ERP pattern features under different experiment conditions

NEMO Functional Ontology

Angela LairdBrainMap

Jessica TurnerBIRNlex

(now part of Neurolex)

CogPO

http://brainmap.org/scribe/index.html

Reconsistituting the ERP domain…

1 sec

TIME SPACE

FUNCTION Modulation of ERP pattern features under different experiment conditions

NEMO ERP Ontology

Observed Pattern = “P100” iff Event type is stimulus AND

FUNCTIONAL Peak latency is between 70 and 140 ms AND

TEMPORAL Scalp region of interest (ROI) is occipital AND SPATIAL Polarity over ROI is positive (>0)

FUNCTION TIME SPACE

PATTERN DEFINITIONS (Revised)

“P100” 1. 70 ms < TI-max ≤ 140 ms2. ROI = Occipital3. IN-mean (ROI) > 0

“N100” 1. 141 ms < TI-max ≤ 220 ms2. ROI = Occipital3. IN-mean (ROI) < 0

“N3c” 1. 221 ms < TI-max ≤ 260 ms2. ROI = Anterior Temporal3. IN-mean (ROI) < 0

“MFN” 1. 261 ms < TI-max ≤ 400 ms2. ROI = Mid Frontal3. IN-mean (ROI) < 0

“P300” 1. 401 ms < TI-max ≤ 600 ms2. ROI = Parietal3. IN-mean (ROI) > 0

SPATIAL TEMPORAL

Cycles of pattern definition, validation, & refinement(MORE ON THIS NEXT TIME…)

Frishkoff, Frank, et al., 2007

Protégé Software for Ontology Development

Overview Agenda

• Introductions & go-to people (7 mins)• Scheduling regular teleconferences (3 mins)• Review of project aims (15 mins)• Contributing to NEMO -- overview (10 mins)

(website, wiki, database)• Overview of current ontologies (25 mins)• Overview of RDF/OWL annotation (Dejing Dou)

43

An Introduction for Annotation• Annotation and Markup

– HTML – XML/RDF/OWL

• Ontology-based Annotation– Ontologies and Data Tables. – Links of Data and Ontological Concepts – Applications

Reference: Siegfried Handschuh, Steffen Staab, Raphael Volz: On deep annotation. WWW 2003: 431-438

44

Annotation and Markup• The idea of Annotation or Markup came from WWW. HTML,

Hypertext Markup Language, is still a well-used markup language. For example, your personal homepage are very possibly written in HTML:

<html> <head> <title>Dejing Dou’s

Homepage</title> </head> <body> …. </body> </html> The tags (annotators) (e.g., title, body..) are well defined and

computer can process and display the text, images …in preferred places, color and font size.

XML/RDF/OWL• The XML, eXtensible Markup Language, lets users self-define new

tags: <?xml version="1.0" encoding='UTF-8'?> <faculty> <name>Dejing Dou</name> <ranking> Assistant Professor

</ranking> <student> Paea Lependu </student> …. </faculty> I defined those new tags (faculty, name, ranking…)

but computer do not know the meaning or the semantics of them.

• Using similar syntax, RDF (Resource Definition Framework) and OWL (Web Ontology Language) allow users to define the semantics of tags as ontologies.

45

A Simple Ontology of University

46

People

Faculty

StaffStudent Assistant Prof.

Associate Prof.

Full Prof.

StringName

Graduate Student Undergraduate

Is_a Is_a

Is_a

Is_a Is_a Is_a

Is_a Is_a

Stringtitle

Numberage

47

Sample Data on the People

School_ID Name Age Title Ranking

950499879

D. Dou 36 Dr. Assistant Professor

950699887

P. LePendu 34 Graduate Student

… … … … …

Data and Ontology

48

School_ID Name Age Title Ranking

950499879 D. Dou 36 Dr. Assistant Professor

950699887 P. LePendu 34 Graduate Student

… … … … …People

Faculty

StaffStudent Assistant

Prof.

Associate Prof.

Full Prof.

String

Graduate Student

Undergraduate

Is_a Is_a

Is_a

Is_aIs_a

Is_a

Is_a Is_a

String title

Numberage

Name

Ontology-based Annotation: the links

49

School_ID Name Age Title Ranking

950499879 D. Dou 36 Dr. Assistant Professor

950699887 P. LePendu 34 Graduate Student

… … … … …People

Faculty

StaffStudent Assistant

Prof.

Associate Prof.

Full Prof.

String

Name

Graduate Student

Undergraduate

Is_a Is_a

Is_a

Is_aIs_a

Is_a

Is_a Is_a

String title

Numberage

Results In RDF/OWL • Computer can process it automatically: <People rdf:ID=“950499879”>

<name>Dejing Dou</name>

<age>36</age>

<title> Dr. </title>

<ranking rdf:resource="#Assistant Professor"/>

</People>

<People rdf:ID=“950699887”>

<name>Paea Lependu</name>

<age>34</age>

<ranking rdf:resource="#Graduate Student"/>

</People>

… 50

What we can do?• Search

– Example: return all data rows related to faculty (i.e., all data of assistant, associate and full professors will be returned.)

• Query– Examples: Give the average age of assistant and associate professors only?What are the difference of age range between faculty and

students? • In NEMO, we will develop ontology-based tools to automatically

answer:Return all PCA factors related to “P100” and “N100” only (Search)

What are the difference of range of time latency between Lab A and Lab B’s “P100” patterns in the same paradigm X ? (Query)

51