Functionality Working Group Report 2010-05-27 Dagobert Soergel University at Buffalo 1.

Functionality Working GroupReport 2010-05-27

Dagobert SoergelUniversity at Buffalo

1

y

2

Scenario 1. Simple solution

• Provider functionality (Web-based)– Find correct town name– Find composite object for that town, show outline– User clicks on outline item to invoke get function for the

component

• Interoperability problems focus on consumer ability to use data formats– cross-domain problem: content - functionality

3

Scenario 1. Simple solution - 2• Consumer needs

– either functions, such as a video editor, that are data-interoperable with the provider data format

– or mediators that can convert (map) from the provider format to the consumer format

• Supported by a function repository that can find the needed functions to run on he consumer's computer or as a Web service– Must be able to deal with needed formats– Must have functionality required by consumer– Must run in consumer's computing environment (functional

interoperabilty)

• Needs function representation language

4

Scenario 2. Simple solution

• DB-mix replicates content from DB A an DB B(establish update schedule)– Cross-domain interperability problem Content –Functionality– DB mix format needs to preserve all information from either format

A or format B, both document structure and metadata, including rights

– Need to convert from format A and format B to format mix. Need mediatorsApplies to document format and metadata format

– Special problem: Mapping thesauri A and B to thesaurs mix

5

Scenario 2. Simple solution - 2

• Search functionality implemented on DB mix– Needs complete feature analysis of search A and search B using

function representation language– Need functions to implement all search features– Can pieces of search A code and search B code be reused?

Question of function interoperability– Also data dependence of search features. Only documents

coming originally from A or B may have data required for a search feature – need to warn userGeneral issue: Warn users of incomplete interoperability

– Look and feel of search interface - product compatibilityMay need to provide several interfaces: native DB mix, A, and BNote: Mix can have A interface for invoking a feature even if the implementation of the feature is different ini mix

6

Scenario 1. Simple solution - 3

• Same problems for document display functionality

7

Function repository

• Enhanced function registries for system interoperability Paper for the ECDL 2010 workshop

• Two-pronged approach– Review existing repositories of Web services and software modules– Specify requirements and vision from WG prespective– Describe how exisiting repositories can be enhanced

• It was noted that a function repositories and software libraries used by programmers are the same thing

• Extended function description template next slides

8

Description / specification /profileof functions

Function Specificationfacilitates the identification of what a function does and how either a system or a human may interact with it.

9

Description / specification /profileof functions - 2

Function description/specification template• The template shown below applies to

– the description of an abstract function, such as search or annotate;– the description of specific software components implementing a

function. – Not all items apply to the general level, or the description stays very

broad, for example with regard to data formats. • Hierarchical inheritance:

The template instance for an abstract function has many subordinate template instances for software components implementing that function

• Two hierarchies of functions: isa and partOf

10

Description / specification /profileof functions - 2

• Template focuses on semantics of function specification• Function description template as a specialization of a

general resource description template (in line with the Reference Model)For example, usual metadata are assumed

• This is a preliminary template. It will be amended as it is applied.

• Template needs to be developed with Architecture WG

11

Function specification template

12

Function Behavior

API/Interface Specification

Dependencies/Relationships/Use

Interoperability Concerns

Assessment. Performance. Advice for use

Usage conditions

Plus all other elements from the general Resource Description Template

Function specification template

13

Function BehaviorDescription: What is doneInteraction with Actors (Systems/Users)

Is the function invoked by the user or the systemWhat actions does the user takeWhat actions does the system takeSpecial user groups /roles; user characteristicsCan the function be applied to different contexts

API/Interface Specification

Dependencies/Relationships/Use

Interoperability Concerns

Assessment. Performance. Advice for use

Usage conditions

Function specification template

14

Function Behavior

API/Interface SpecificationInput: Data and parameters, data formats / standardsOutput: Data and parameters, data formats / standardsPreconditionsPostconditions

Dependencies/Relationships/UseOperating environment in which the function runs. Other functions it needsOther functions that invoke this functionOther functions invoked. Composite functionsWork flow

Interoperability Concerns

Assessment. Performance. Advice for use

Usage conditions

Function specification template

15

Function Behavior

API/Interface Specification

Dependencies/Relationships/Use

Interoperability ConcernsWhat is required for interoperability (distinguish type of interoperability, for example product compatibility).How does a specific implementation meet these requirements

Assessment. Performance. Advice for useFor specific types of functions, such as search, the template should include quality parameters (for example from the DLRM to assist with assessement and facilitate comparison of assessments – software evaluation criteria

Usage conditions

General dl.org issues• The Reference Model has not taken hold

– Partially a problem of presentation– Reference model can serve as the beginning of a detailed

function ontology needed for function representation

• Relationship between the Reference Model, the State-of-the-Art report, and the Cookbook needs to be clarified. – The three documents need to be seen as three sides of the same

pyramid and strongly cross-referenced

• Cookbook needs to have best practices, need input from practice

• All documents need to be edited by a an expert who is a native English speaker

• All documents need to be live, Wikipedia-style (with stronger central editorial control)

16

General dl.org issues - 2

• All domains need to specify – requirements for data formats to capture data in that domain, such

as rules that codify policies– requirements for functions that are needed to process these data for

domain-specific purposes, such as the interpretation of rules to enforce policies

• All domains need ontologies

• Bilateral interoperability is not enough, need multilateral and union solutions

• Provider – consumer is not the best terminology. Often there is interchange rather than one-way flow

17

WG Functionality

• Other members

• Questions

18

Scenario 1. Simple solution

• S

19

Ultimate objectives• Promote rich functionality over a wide range

of systems with a consistent interface • Promote best practices and innovation by

educating DL designers, developers, administrators, and users about the rich array of DL functionality

• Enable finding and reusing software modules that implement desired functionality– for developers: reuse existing modules and design for interoperability– for DL managers:cutting-edge functionality in configuring a DL system– for users: run a module "on the fly" to accomplish a task.

• Enable federated search

20

Outline

• Objectives and scope• Function Interoperability Framework

– Definition of function– E-R schema for database of function descriptions– Definition of function interoperability– Function description template

• Conclusion

21

Objectives and Scope

22

Ultimate objectives• Promote rich functionality over a wide range

of systems with a consistent interface • Promote best practices and innovation by

educating DL designers, developers, administrators, and users about the rich array of DL functionality

• Enable finding and reusing software modules that implement desired functionality– for developers: reuse existing modules and design for interoperability– for DL managers:cutting-edge functionality in configuring a DL system– for users: run a module "on the fly" to accomplish a task.

• Enable federated search

23

Interoperability and reuse scenariosScenarios1 Find desired functions, and modules that implement them, and

assess their interoperability. Enable functionality sharing2 Enable content sharing and federated search3 Make switching from one DL to another easy for the user

Dealing with these scenarios requires1 Understanding the many ways in which functions interoperate2 A database with detailed descriptions of functions,

revising and extending the DELOS Digital Library Reference Model

Solution: Function Interoperability Framework

24

Interoperability and reuse scenariosExamples

• The developer of a Browse module looks for anautomatic clustering module to incorporate browsing by cluster

• A DL administrator wants to make available a better image search system

• A user found 30 documents in a DL.Wants to invoke a Web service to create a multi-document summary

25

The role of the WG Functionality

26

27

Function Interoperability Framework

28

What is a function?

A function in the DLRM is an action a DL component or a DL user performs (abstract function, emphasis on semantics, what does the function do).

In the followingfunction is used broadly to mean either

abstract function or software component implementing a function

29

Some Functions illustrating Interoperability issues

30

Behind the scene For usersFeature extraction

Classification / clustering

Sharing authority files

Log file analysis

Sharing user profiles

Harvesting , aggregating

Shared storage and backup

Federated search

Incorporating content from other places on the fly

Display and visualization

Timelines

Maps

Playing videos

Same look-and-feel browse

Three parts of the Function Interoperability Framework

• An entity-relationship schema• A taxonomy of ways in which functions can

interoperate• A template for the description of functions

and software components

Note: Strong overlap with Architecture WG

31

What is function interoperability 1

1 Interoperability (system perspective, focus on software components)1.1 Composability (f2 can work with f1)1.2 Replaceability / interchangeability (f2 can replace f1)

(f1 and f2 serve same purpose)1.1 and 1.2 can be based on process or on data (exchanging data or using same data)

2 Cross-function (cross-product ) compatibility (user perspective)Similar detailed functionality and user interface

32

Description / specification /profileof functions

Function Specification: facilitates the identification of what a function does and how one (either a system or a human) may interact with it.

Function description/specification template• The template shown below applies to

– the description of a general function, such as search or annotate;– the description of specific software components implementing a

function. – Not all items apply to the general level, or the description stays very

broad, for example with regard to data formats.

• Template focuses on semantics of function specification• This is a preliminary template. It will be amended as it is applied.

33

Function specification template

34

Function BehaviorDescription: What is doneInteraction with Actors (Systems/Users)

Is the function invoked by the user or the systemWhat actions does the user takeWhat actions does the system takeSpecial user groups /roles; user characteristicsCan the function be applied to different contexts

API/Interface Specification

Dependencies/Relationships/Use

Interoperability Concerns

Emerging function ontology

Ontology of functionsFunction specification vocabulary

Will emerge over time as the database of function descriptions is populated through wide collaboration (crowd-sourcing)

35

Sub-functions of annotate

36

Select object to be annotated(need to indicate selection method)

Mark region in the object(many different methods depending on the object)

Select type of annotation (highlight, mark with special meaning, text, image, sound)

If text, image, sound

Specify relationship to object to be annotated

Select or create the annotating object (possibly specifying a region

Annotating within one system

Annotating across systems

Conclusion

37

Recap

• ObjectivesInteroperability use cases

• WG will produce a Function Interoperability Framework

• dl.org should set up an environment in which the DL community can produce a database of function descriptions

38

Expected Outcomes

• Interoperability State-of-the-Art survey• Extensions to the Delos Reference Model• Contributions to the Cookbook:

Best practices in functionalityIdea of design patterns for DLs

• One or more papers• Training course materials

WG Activities

• Definition of Functionality WG scope• Specification of Function Interoperability• Contribution to the State-of-the-Art Survey

– Function Interoperability Issues and Solutions

• Contribution to the Reference Model– Extensions to facilitate additional topics e.g. composite

functions

WG Activities cont.

• Contribution to the Cookbook– Interoperability Issues and Solutions

• Drafting of a “Functionality Specification Template”• Specification of related Use-Cases

– Integration of D4Science and DRIVER platforms

• Presentation of WG outcomes at the 1st DL.org Workshop in Corfu (ECDL 2009)

WG Activities cont.

• Publications– “A Functionality Perspective on Digital Library

Interoperability”, to be presented in ECDL 2010

• WG Meetings– Three (3) face-to-face meetings– Four (4) virtual meetings

Thank you

Interoperability use casesFocusing on behind-the-scenes operations• Centralized services that provide the same service to

multiple libraries – classification servers– format conversion– data validation

• The services must be interoperable − based on data − with many DLs

• Can be achieved through– standards for data formats – standardized authority files.

44

What is function interoperability 2

A Interoperability of functions based on process

A1 Interoperability of use (composability)Function f1 can use function f2 (conversely, f2 can work in the

framework of f1))

A2 Special case: Interoperability with environment E (composability)Function f1 can work in environment E

A3 Interoperability based on working in same operating environment E (replaceability / interchangeability) If Function f1 can operate in environment E AND f2 (having the same purpose as f1) can also operate in E, Then f2 can replace f1

45

What is function interoperability 3B Interoperability of functions based on data (content)

B1 Interoperability based on exchanging data (composability)If Function f1 can operate on the output of f2, Then f1 and f2 can work together. f1 and f2 may also exchange data as they run concurrently

B2 Interoperability of functions with data (composability) If f1 can make use of data set D or of data formatted according to DFThen Function f1 is interoperable with a data set D or a data format DF

B3 Interoperability of function based on using same data (replaceability)If Function f1 is interoperable with a data set D or data format DFAND Function f2 (serving the same purpose as f1)is also interoperable with D or DF, respectively,Then f2 can replace f1

46

Issues in interoperability

• API mismatch• Mismatch in programming environments

Needed components missing• Mismatch in data formats

(overlap with WG Content)

47

Interoperability use cases

Exchange of program modules between D4 Science and Driver•Each system would describe the functions it implements (e.g. feature extraction from documents or data transformation using grid resources), considering

– the semantics of the function (what the program module can do) – the technical (and, as relevant, administrative) conditions of use.

•Each system could then search the functions offered by the other and reuse program modules.

48

Interoperability use cases

Single depositEuropean project OpenAIRE

– Central portal where users come to deposit their publications.

– The internal deposition service subsequently forwards/deposits them in the corresponding local repositories.

– Requires interoperable functionalities among the various repository platforms.

49

50

Browsing Collaborating Customizing Filtering Providing access Recommending Requesting Searching Visualizing

Annotating Classifying Clustering Evaluating Extracting Indexing

Measuring Publicizing

Rating Reviewing (peer)

Surveying Translating

(language)

Conserving Converting

Copying/Replicating Emulating Renewing

Translating (format)

Acquiring Cataloging

Crawling (focused) Describing Digitizing

Federating Harvesting Purchasing Submitting

Preservational Creational

Add Value

Repository-Building

Information Satisfaction

Services

Infrastructure Services

E-R schemafor a function description database

Entity types (resource types) (examples)

Relationship types (examples)

Function

SoftwareComponent(a software system, software module, or code snippet)

DesignPattern (Rike Brecht, Doct. Cons.)

Data set

Data format

Resource <hasComponent> Resource

Function <implementedBy> SoftwareCo.

Function <represented by>DesignPattern

Resource <interoperableWith> Resource

51

Function specification template

52

Function Behavior

API/Interface SpecificationInput: Data and parameters, data formats / standardsOutput: Data and parameters, data formats / standardsPreconditionsPostconditions

Dependencies/Relationships/UseOperating environment in which the function runs. Other functions it needsOther functions that invoke this functionOther functions invoked. Composite functionsWork flow

Interoperability ConcernsWhat is required for interoperability (distinguish type of

interoperability, for example product compatibility).How does a specific implementation meet these

requirements

Sub-functions of search

53

Quick Search Advanced SearchEnter a query and click search

Enter keywords or phrases for selected field

Limit results to

Search subscribed titels

Clear

Enter a query and click search

Enter keywords or phrases for selected fields

Select keyword from a list

Select Boolean operator (explicit)

Define phrase match (explicit)

Clear

Search within results

Limit results to (preselection)

Sort by (preselection)

Select display options

Display X results per page

Display search history

y

54