On Path-Centric Navigation and Search Techniques for Personal - - PowerPoint PPT Presentation

On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

2 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Introduction Preconditions and the Model Navigation inside Topic Maps Search based on Topic Selection

Outline

3 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Introduction

Which problem do you address?

4 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Viele Informationen, Programme und Geräte

DB DMS

??

Daily challenges with multiple devices, various tools and different location of data

5 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

DB DMS

Collect and interconnect data, to ease daily work with information

6 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

MIDMAY autonomously creates topic maps from data sources and preserves the user given structures

7 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Preconditions and the used Model

What is the foundation of your approach?

8 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Topic map is designed to find the desired reference by leveraging redundancies in data sources Global typing schema across all extractors (PSIs, PSIDs) Association type reflects and unifies semantic

• f a property, type or hierarchical relation

No directionality inherent in an association Each entry unique in the knowledge space

• f a user -> consistency

9 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Graph Definition Topic map graph G is described by the pair (V, E). V is finite set of vertices mapped to topics E is a binary relation on V, representing the undirected associations between the topics. Additionally, E explicitly contains the binary relations between topics and their types => G contains a vertex in V for every type topic. Each edge (vi, vj) œ E is given a constant configurable weight wij depending on the type of association and the search mode

10 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Navigation inside Topic Maps

What do you mean with path centric?

11 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Cycling through the Graph

Term Selection Topic Selection Association Selection

navigate navigate redo redo redo mark

12 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Points of Entrance & Navigation Aid Type Lists

13 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Points of Entrance & Navigation Aid Hierarchy Root

14 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Topic Paths

Graz [Location] Meeting E Example Ltd. d. [Event] John D Doe

• e

[Perso son] Discussed Pape per [Email] Semantic Distanc nce.ppt [Att ttachment] PPT [Filetype]

15 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Path Navigation Follow paths in mind

I want to find a document, but can’t remember some fitting keywords However, I recall that the document was sent by someone I meet in a meeting in Berlin Let’s start the search with an item I can name: Berlin …

16 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Path Navigation Screenshots (1)

17 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Path Navigation Screenshots (2)

18 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Path Navigation Screenshots (3)

19 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Path Navigation Screenshots (4)

20 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Search based on Topic Selection

How does selecting topics help us searching?

21 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Selecting topics to specify Search Query Select two topics ta, tb and choose mode (0-3) weighted Breadth-first Search from ta and tb, until the path with the lowest value is found next path by removing the edge that connects both waves -> sufficient if all topics in possible paths are presented at least once

22 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Example: Show path between the topics Graz and PPT

23 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Path Maths Definitions introducing Bit Vectors The vector Bp

ab indicates the presence or absence of

topics <t1, t2, t3, ..., tk> in path p between topic ta and tb. ( Bab : topic presence for all shortest p)

24 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Marking multiple topics to calculate a set of relevant result topics

• Search Modes
• Mode 0, the default mode that only uses the

structure structure of the topic map

• Mode 1, to focus the search on topics equally

connected by hierarchy hierarchy

• Mode 2, to focus the search on equal properties

properties of marked topics

• Mode 3, to focus the search on equal types

types of marked topics

25 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Example: Search all employees involved in project MIDMAY which authored a PPT presentation Think of topics related to search problem

I’m looking for a person person .. He’s involved in project MIDMAY MIDMAY The file type is PPT PPT

Navigate to the topics and mark them Start query

26 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Search – Add MIDMAY Topic

27 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Search – Add PPT Topic

28 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Conclusion

What’s the benefit and what are the remaining challenges?

29 Jens Heider Julian Schütte TMRA ´07 – On Path-Centric Navigation and Search Techniques for Personal Knowledge Stored in Topic Maps

Already existing data can be used to offer an intuitive way to search for information Path and set calculation provide search functionality in topic maps beyond keyword search techniques for non-technical users Challenges

• Capacity of Topic Maps Engine
• Enhanced UI for query, bringing the full flexibility of

path calculation to the user

Topic Maps can help to tackle the daily work with stored information