Inter-context Trust Bootstrapping for Mobile Content Sharing

Post on 21-Jun-2015

1,209 views 1 download

Tags:

description

This talk will look at how a trust model running on device A determines the extent to which A should initially trust device B in a given context (content category). It does so by considering two cases: in the first, A does not know B at all; in the second case, A knows B but in contexts other than that of interest. For each of those two cases, this talk will discuss the most recent proposal that improves on existing solutions (TRULLO and distributed propagation), and will also attempt to suggest new research directions (such as private collaborative filtering - post & more).

transcript

Inter-Context Trust Bootstrapping

for Mobile Content Sharing

(daniele quercia)(stephen hailes & licia capra)

What do I do?

Research @

what I research?

Reputation Systems for Mobiles

What’s that?

Example:antique markets

Problem: Visitors cannotsee prices of everything!

Solution: Sellers disseminate

e-ads, and visitors collect them

Problem: Sellers may disseminate irrelevant ads

Proposal:

They may keep track of

which sellers send

irrelevant ads

Daniele Quercia

Trust model on A: how A decideswhether to rely on B

to visit a stall

Daniele Quercia

To decide whether to rely on B, A has to

set its initial trust in B

Daniele Quercia

3 Existing Solutions

Daniele Quercia

1. Fixed values ( over-simplified)

Daniele Quercia

2. Recommendations

( fake ones)

Daniele Quercia

3. Similar contexts( universal ontology)

Daniele Quercia

Two cases: B is

1. unknown 2. partly known

Daniele Quercia

1. B is unknown

Daniele Quercia

Popular way:

Trust propagation (transitivity)

?A B

C

Daniele Quercia

Meant for the Web & Proved on “binary” ratings

Daniele Quercia

Algorithm rating • unrated trust relationships (needed)

1

?A B

C2

1 2

?• unrated nodes (chosen)

AB

AC CB

?

Idea:1. Similar nodes together2. Find function:•same ratings for rated nodes•similar ratings for neighbours

Daniele Quercia

Tested on real data (Advogato: > 55K user ratings)

Daniele Quercia

2. B is partly known

Daniele Quercia

Popular way:

Inter-context Lifting

Greek Coins

Roman Coins

Coins Chairs

Antiques

Daniele Quercia

Idea: Users …

> Don’t share ontology

> Extract “features” from their own ratings

Daniele Quercia

Idea: Users …

> Don’t share ontology

> Extract “features” from their own ratings

Daniele Quercia

How to extract?

Daniele Quercia

Singular

Value

Decomposition

Daniele Quercia

Beauty: features not user-specified

BUT learnt

Daniele Quercia

Tested on simulation with real parameters

Daniele Quercia

Tested on Nokia 3230

Max: 3.2 ms !

Daniele Quercia

What I’ve told you is on “mobblog UCL” (google it)under tag: “bootstrapping”

Daniele Quercia

Daniele Quercia

And User Privacy?

Daniele Quercia

Private filtering(Google for “mobblog private filtering”)

Daniele Quercia

And Resource Discovery?

Daniele Quercia

Folksonomy for mobiles

Daniele Quercia

And Attacks?

Daniele Quercia

Further Research(join mobblog !)