+ All Categories
Home > Documents > Daedalus

Daedalus

Date post: 24-May-2015
Category:
Upload: roberto-trasarti
View: 715 times
Download: 1 times
Share this document with a friend
Popular Tags:
12
R. Ortale (2), E. Ritacco (2), N. Pelekis (3) R. Trasarti (1), F. Giannotti (1), C. Renso (1), G. Costa (2), G. Manco (2), Y. Theodoridis (3) (1) ISTI-CNR, Pisa, Italy (2) ICAR-CNR, Rende (CS), Italy (3) Univerity of Pireus, Athens, Greece The DAEDALUS Framework: Progressive Querying and Mining of Movement Data
Transcript
Page 1: Daedalus

R. Ortale (2), E. Ritacco (2), N. Pelekis (3)

R. Trasarti (1), F. Giannotti (1), C. Renso (1), G. Costa (2), G. Manco (2), Y. Theodoridis (3)

(1) ISTI-CNR, Pisa, Italy(2) ICAR-CNR, Rende (CS), Italy

(3) Univerity of Pireus, Athens, Greece

The DAEDALUS Framework:

Progressive Querying and Mining of Movement Data

Page 2: Daedalus

2

Motivation

Knowledge discovery is a multi-step process, that involves data preprocessing, pattern mining stages and pattern postprocessing.

Lack of a unifying framework, where mining tools are specific components of the knowledge discovery process.

Page 3: Daedalus

3

Motivation

Which trajectories support T-pattern that are inside a polluted area?

SELECT Trajectories.id

FROM Patterns, Trajectories, Polluted_Areas

WHERE Trajectories.object satisfies Patterns.object

AND Polluted_Areas.geometry includes Patterns.object. Geometry

This is an example of Join query between patterns, trajectories and background geographic knowledge

Page 4: Daedalus

4

Motivation

Amalgamating elements from different worlds causes an impedence mismatch

Different representations, different objectives

Idea

Explicitly represent objects in these different worlds

Provide bridges through the worlds

Page 5: Daedalus

5

The Two Worlds framework

Filtering operators: manipulate basic objects.

Mining operators: extract properties from samples. K:DM

Population operators: detect samples exhibiting properties. P:DxMD

Page 6: Daedalus

6

From Two Worlds to DaedalusHermes is the repository of both data and models.

Hermes has been extended to represent objects in M-World: Model_TAS, (T-Pattern)

The mining operator is realized by calling an external algorithm.

The populate operator has been defined on Hermes

Page 7: Daedalus

7

The Data world

Represents the entities to be analyzed, their properties and mutual relationship

Our context: trajectory data

Example:

TABLE Trajectories

ID : integer

Type : {vehicle, pedestrian}

Object : Moving_Point

Page 8: Daedalus

8

The Data World – Data filtering

SELECT t FROM Trajectories t WHERE t.type=“veichle”

SELECT count(t) FROM Trajectories t , Polluted a WHERE t.object intersects a.geometry

SELECT count(t) FROM Trajectories t , RushHours r WHERE t.object at_period r.period

SELECT count(t) FROM Trajectories t , Trajectories y WHERE t.object intersects y.object and y.id=3

Page 9: Daedalus

9

Model representation

For T-Pattern, a Model_Tas is defined in Hermes as:

Sequence of <Region, <Minimum travel time, Maximum travel time>>

Model_TAS: VARRAY <SDO_Geometry, <TAU_TLL.interval, TAU_TLL.interval>>

<A,<10,30>; B<5,60>; C<nd,nd>>

A

B

c

10,305,60

Page 10: Daedalus

10

The Daedalus system

DAEDALUS provides a Data Mining Query Language based on SQL, that includes basic mechanisms for interactive queries on D-World and M-World

Page 11: Daedalus

11

The Daedalus System Architecture

HERMES

Model_TASPackage

MODMediator

Controller

Parser

ObjectTranslator

MiningEngine

T-PatternAlgorithm

User Interface

TASTranslation

Library

Moving_pointTranslation

Library

ObjectStore

DMQL query

Page 12: Daedalus

12

Demo

We will show the Daedalus prototype

It has been developed in Java, based on Hermes and plugged with T-Pattern and clustering algorithms.

We will give some query examples to show the expressiveness of the language.


Recommended