sbsgrid visual linked-data_publisher_explained

12
SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 1 / 12 © SBSGRID.NET Visual Linked-Data Publisher Explained Publishing database tables via triples into Linked-Data for instant semantic graph-traversal and search/query functionality (Dataset example uses Artists, Galleries and Artworks) Contents 1. Introduction 2 2. Characteristics 2 3. Publishing Patterns 3 3.1. Master-Detail Relationship 3 3.2. Link-Table Relationship 4 3.3. Parent-Child Relationship 5 3.4. Complex Relationship 6 4. Semantic SearchQueries 7 4.1. Introduction 7 4.2. Galleries in New York City showing Modern Paintings 9 4.3. Asian Art within a Price Range 10 4.1. Gerhard Richter's Exhibitions in Germany 11 4.1. Paris Context Map 12

Upload: bernardangerer

Post on 29-Nov-2014

500 views

Category:

Documents


0 download

DESCRIPTION

 

TRANSCRIPT

Page 1: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 1 / 12

© SBSGRID.NET

Visual Linked-Data Publisher

Explained Publishing database tables via triples into Linked-Data for instant semantic graph-traversal and

search/query functionality

(Dataset example uses Artists, Galleries and Artworks)

Contents

1. Introduction 2

2. Characteristics 2

3. Publishing Patterns 3

3.1. Master-Detail Relationship 3 3.2. Link-Table Relationship 4 3.3. Parent-Child Relationship 5 3.4. Complex Relationship 6

4. Semantic SearchQueries 7

4.1. Introduction 7 4.2. Galleries in New York City showing Modern Paintings 9 4.3. Asian Art within a Price Range 10 4.1. Gerhard Richter's Exhibitions in Germany 11 4.1. Paris Context Map 12

Page 2: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 2 / 12

© SBSGRID.NET

1. Introduction

SBSGRID has an "instant use" approach to its Linked-Data1 management. A graphical user interface allows to select the

tables and relationships to be published (when using relational databases as data-sources). A generic SearchQuery

interface instantly and incrementally allows for the execution of semantic search, query and graph browsing (superset of

SPARQL functionality). The development of a middleware is not necessary. Namespace alignment (through the

submission of additional facts like sameAs, subClass or hasRole) can be done within the SearchQuery interface (during

run-time when the application is already deployed).

2. Characteristics

Procedure is ad-hoc and incremental (no definition of XML schemas necessary). Semantic SearchQueries can be

executed instantly.

Procedure is visual. Tables and columns are dragged and dropped. Relationships between tables are drawn

through lines.

Works for all JDBC/ODBC compliant data-sources.

Necessary SQL is created in the background.

Necessary network data transport protocol statements (triple strings) are created in the background.

Additional SameAs- or SubClass-statements (namespace alignment) are submitted within the SearchQuery

interface.

The import via 'triple strings' is a lean and easy way of semantic database publishing. However existing tools (like

D2R) which use RDF/XML (R2RML) can be used just as well.

1 see http://www.linkeddata.org

Page 3: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 3 / 12

© SBSGRID.NET

3. Publishing Patterns

3.1. Master-Detail Relationship

In this example an artist has a nationality assigned from a lookup-table.

The artists and the nationality tables are published

The join between the two nationality-ID fields is created through the blue line

The name column is published as "Artist Name" and "Nation Name" respectively

Triple Stream Result

Database Tables, Views

and Columns (Drag &

Drop)

Published Tables,

Relationship and

Columns

Page 4: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 4 / 12

© SBSGRID.NET

3.2. Link-Table Relationship

In this example artists are linked to galleries through a relationship table.

The artists and the galleries tables are published

The join between the artist-ID and gallery-ID field is created through drag and drop (blue line property window)

The relationship is published as "Artist Gallery" (blue line property window)

Page 5: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 5 / 12

© SBSGRID.NET

3.3. Parent-Child Relationship

In this example styles are defined in a parent/child tree structure.

The tags table is published and linked to itself

The join between the parent-ID and the tag-ID field is created through the blue line

The relationship is published as "SUBCLASS" (blue line property window)

Page 6: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 6 / 12

© SBSGRID.NET

3.4. Complex Relationship

In this example artists are linked to galleries through a relationship table. However this relationship table has additional

fields.

The artists and the galleries tables are published

The relationship table is published as a Role. The SearchQuery reasoner will identify this role as a complex

relationship and handle it accordingly (polymorphic roles).

The join between the artist-ID and gallery-ID field is created through drag and drop (blue line property window)

Note: These polymorphic roles resolve complex relationships (e.g. the same person has relationships with different

organizations in different roles based on different works and activities) in a declarative and automatic manner. The roles

adapt automatically (the resultset changes) based on the given context in the query. Thus the logic is not hard-coded in

any SQL-statement or middleware code.

Page 7: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 7 / 12

© SBSGRID.NET

4. Semantic SearchQueries

4.1. Introduction

Information stored in database tables. Data are not linked. Contextual and associative questions can not be asked:

Information stored as Linked-Data. Contextual and associative questions can be asked:

Page 8: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 8 / 12

© SBSGRID.NET

Keyword are semantically identified and the answer is derived from the information-network.

Example query on "Galleries in New York City showing Modern Paintings":

Note: The following screenshots show a generic (!) way to query and traverse the information graph. Customization

(look & feel) of the user interface can be quickly changed through the underlying XML/REST or JSON interface.

Page 9: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 9 / 12

© SBSGRID.NET

4.2. Galleries in New York City showing Modern Paintings

Semantic reasoning trace:

The system returns the gallery “Claire Oliver” (and others). The following reasoning logic led to this result:

the gallery “Claire Oliver” is linked to "Manhattan" and "Gallery"

the artist Helen Frankenthaler is linked to "Claire Oliver" and "Abstract Expressionism"

"Abstract Expressionism" is a subcategory of "Modern Painting" (reserved predicate SUBCLASS)

"Manhattan" is a subcategory of "New York City" (reserved predicate SUBCLASS)

Semantic identification

of the category "Modern

Painting"

Semantic filtering of the Modern

Painting information-space to galleries

in NYC

Additional related results:

Artworks which are linked

to the galleries

Page 10: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 10 / 12

© SBSGRID.NET

the semantic space makes the connection and "infers" that "Claire Oliver is a [GALLERY] which shows

[MODERN PAINTINGS] in [NEW YORK CITY]"

Note: Semantic search solutions work on the concept level not data level (no Linked-Data). They have an ontological

body put on top of a fulltext-search index. This means that the information that Abstract Expressionsim is a subclass of

Modern Painting is in the system. However the relationship between Claire Oliver and Helen Frankenthaler lies within

the data and is not covered by the ontology. Thus the inference/reasoning from above can not conducted.

4.3. Asian Art within a Price Range

Semantic Suggest-List Numbers identified as

artwork prices in the given

context

List of countries

Page 11: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 11 / 12

© SBSGRID.NET

4.1. Gerhard Richter's Exhibitions in Germany

Additional related results:

Galleries which are linked

to the exhibitions

Semantic identification

of the string "Gerhard

Richter"

Semantic filtering of Richter's

information-space to

exhibitions in germany

Page 12: Sbsgrid visual linked-data_publisher_explained

SBSGRID_Visual_Linked-Data_Publisher_Explained.docx 12 / 12

© SBSGRID.NET

4.1. Paris Context Map

During typing a context map is

displayed