Knowledge-Based Semantification of Business Communications in ERP Environments

thickbugSoftware and s/w Development

Oct 28, 2013 (3 years and 11 months ago)

110 views

Knowledge
-
Based
Semantification

of Business
Communications in ERP Environments





Marios

Meimaris

and
Michalis

Vafopoulos



MBA Thesis

National Technical University of Athens

University of Piraeus

2012


1
st

International Workshop on
Engineering the Semantic Enterprise (ESE2012)

The
13
th

International Conference on Web Information System Engineering (WISE 2012)

New Business Intelligence


Information is increasingly
dynamic

and
heterogeneous


Information flows are constantly changing form and direction


Faster lifecycles of data


Within corporate contexts:


Internal sources
(
ERP databases, company knowledge,
mobile phones
etc.)


External sources
(
Web 2.0, Semantic Web services, (Linked) Open Data
etc.)

National Technical University of Athens
-

University of Piraeus

2

External Sources


Social Media

3

National Technical University of Athens
-

University of Piraeus


Rich

dynamic

layers

of

information,

both

inbound

and

outbound


Social

media

sales

>
$
30
bn

by

2016
,

>
50
%

of

web

sales


~
-
60
%

in

marketing

costs

with

the

use

of

social

media

in

2012


29
%

of

brands

have

used

social

networking

sites

to

promote

offers


Inefficiency

of

social

media

monitoring

solutions
,

ignored

and

unexploited

(e
.
g
.

70
%

of

product/service

complaints

are

ignored

by

businesses)


*Source: https://www.gartner.com/it/page.jsp?id=2092217

External Sources


(Linked) Open Data


Open

Data
:

the

notion

that

certain

kinds

of

data

should

be

freely

accessible

on

the

web
.


Linked

Data

:

representation

as

resource

nodes

that

are

interlinked

with

property

edges

that

connect

different

datasets
.


No

redundant

data

in

companies’

internal

databases

=>

no

management

costs
.



No

need

to

keep

data

updated

=>

no

curation

costs
.


Retrieval

is

easy

to

perform

on
-
demand
.


e
.
g
.

OpenCorporates
.
com

provides

open

up
-
to
-
date

information

about

international

corporate

entities

in

machine
-
understandable,

open

formats

(XML,

RDF,

JSON)



National Technical University of Athens
-

University of Piraeus

4

External Sources


(Linked) Open Data



National Technical University of Athens
-

University of Piraeus

5

Source: http://dbpedi a.org

External Sources


Semantic Web


Organizations

adopt

Semantic

Web

standards

and

knowledge

representation

formalisms

in

pursuit

of

knowledge
-
based

collaboration
.


Builds

bridges

between

remote

and

heterogeneous

resources
.


Ontologies

provide

common

semantic

vocabularies,

this

way

promoting

interoperability

between

autonomous

software

agents
.



E
.
g
.

GoodRelations

e
-
commerce

ontology



Google
,

Yahoo!
,

BuyMore
,

OpenLink

Software

(among

>
10
,
000

businesses)
.


Publish

or

harvest

offers

described

with

GoodRelations

terms


F
ind

related

products/services,

identify

competing

companies,

strategies

etc
.




National Technical University of Athens
-

University of Piraeus

6

Internal Sources


ERP Database


Enterprise

Resource

Planning

(ERP)

systems

are

widely

used

as

corporate

software

solutions
.


Customer

Relationship

Management

(CRM),

Human

Resources

(HR),

Logistics

etc
.
,

limited

data

management
.


Forms

the

information

core

of

a

business


Our

implementation

builds

on

OpenERP
,

a

widely

used

opensource

ERP

system

with

rounded

capabilities
.

National Technical University of Athens
-

University of Piraeus

7

Internal Sources


Mobile Devices


Smartphones

as

data/

metadata

live

gathering

points
.



Aggregating

and

processing

the

gathered

data

recreates

usage

contexts

in

communication

environments
.


The

Android

environment

is

equipped

with

an

extensive

SDK

that

makes

the

above

possible
.


Within

companies,

communication

networks

share

a

common

layer

of

data

exchange
.

The

use

of

smartphones

makes

this

layer

identifiable

and

quantifiable
.


Three

levels

of

information

retrieval
:

National Technical University of Athens
-

University of Piraeus

8

Internal Sources


Mobile Devices


Communication

Level

information

exhibits

a

high

degree

of

dynamicity
,

vs
.

relatively

static

Contacts

Level

and

ERP

database
.


F
rom

voice

calls

and

text

messages

(
sms
,

emails,

IM

etc
.
)

to

calendar

event

participations

and

social

networking

interactions
.


Within

the

company

AND

between

the

company

and

the

outside

world

=>

network

and

sub
-
networks

=>

reflection

of

actors’

behaviours


Rich

taxonomy

in

the

communications

domain

=>

diverse

classifications
.


Aggregation

of

all

communication

instances

creates

a

descriptive

communication

network
.



Existing

actor

(employee,

client

etc
.
)

profiles

are

enriched

with

communication

information

on

the

fly

=>



pattern

analysis

of

communication

histories


application

of

informed

communication

strategies

in

future

endeavours
.



National Technical University of Athens
-

University of Piraeus

9

Knitting it all together

National Technical University of Athens
-

University of Piraeus

10

Knitting it all together


Data

diversity

with

respect

to

form,

structure,

provenance

and

scope
.


The

need

of

a

common

substrate

leads

us

to

the

use

of

ontologies

in

all

levels

(KM

and

SW)
.

Benefits
:


Better

knowledge

management
,

semantic

structure

and

machine
-
understandable

meaning
.

Rule

based

reasoning,

intelligent

querying,

interchangeable

knowledge
.


Data
-
sharing

and

semantic

web

interoperation

with

others,

through

the

use

of

reference

ontologies

and

SW

services
.



Reuse

of

existing

ontologies

for

the

data

that

is

designed

to

be

shared


We

created

an

ontology

to

represent

communications
,

tasks

and

mobile

usage

context
.

National Technical University of Athens
-

University of Piraeus

11

Existing
Ontologies


Friend
-
of
-
a
-
Friend

(FOAF)
:

Describes

people

and

organizations


GoodRelations
:

Describes

e
-
commerce

offers


SEMA
4
A
:

Contains

basic

terminology

for

describing

threats

and

alerts


Dublin

Core
:

Provides

vocabulary

for

describing

non
-
functional

properties

of

resources


National Technical University of Athens
-

University of Piraeus

12

Local Ontology


Main

contribution
:

Communications

domain

description


Fills

the

gaps

between

ontologies


Provides

(domain
-
specific)

descriptions

for

the

Company,

such

as

:


Internal

Structure


Employee

and

Department

Hierarchy


Classifications

of

resources

for

internal

reference


National Technical University of Athens
-

University of Piraeus

13

Scenario


We

have

applied

the

implementation

to

a

threat
-
handling

scenario

that

utilizes

the

above

design
.



The

implemented

system

is

mostly

concerned

with

internal

sources


Components

that

interact

with

external

sources

(except

OpenCorporates
)

have

not

been

implemented
.


National Technical University of Athens
-

University of Piraeus

14

Scenario


The system notifies with
alerts

when external information triggers threats
about offers of particular products/services. The implemented system
handles the alerts in two ways:


Automatic creation of suggestions
for handling a particular threat:


I
dentification of subsets of employees
that will handle the threat,


A
ssignment of particular clients
(or sets of clients) to these employees,


S
uggestion of
dominant communication strategies
(by pointing out communication
styles).


Browsing environment with related knowledge
and statistics for the
particular threat, in order to assist managers in manually assigning
threat handlers.


National Technical University of Athens
-

University of Piraeus

15

Scenario


Past information drawn from Contacts and Communications Levels shared
among salespeople within the company, based on accessibility rules.


Overall context of the scenario is derived from the ERP system


The system automatically classifies market and clients in order to outline
the conditions


Results are made available to the salesperson and/or their manager(s)
w.r.t. ontology rules


Salespeople adapt their operational plan and
communucation

strategy
according to the results


The results of their modified strategies are quantified and measured
before they are made part of the shared knowledge, thus rebooting the
cycle


National Technical University of Athens
-

University of Piraeus

16

Threat Handling

National Technical University of Athens
-

University of Piraeus

17

Technical Details


Programming Languages and Environments


Java


Android SDK


Semantic Data Management


Jena Framework (RDF conversion)


RDF (data storage)


OWL (ontology)


SPARQL (querying language)


OpenLink

Virtuoso (quad store)


Web Environment


PHP


JSP (
JavaServer

Pages)


National Technical University of Athens
-

University of Piraeus

18

Data Extraction and Conversion

National Technical University of Athens
-

University of Piraeus

19


ERP data are extracted, converted and
stored in the Virtuoso quad store as shown

Data Extraction and Conversion


Android data are
retrieved,
converted and
stored in the
quad store as
shown in the
first figure


Direct contact
data exchange
from ERP to
Android is
shown in the
second figure


Common
contacts level,
synchronization
of all platforms
based on their
role/permissions

National Technical University of Athens
-

University of Piraeus

20

Screenshots

National Technical University of Athens
-

University of Piraeus

21

Screenshots

National Technical University of Athens
-

University of Piraeus

22

Screenshots

National Technical University of Athens
-

University of Piraeus

23

Screenshots

National Technical University of Athens
-

University of Piraeus

24

Screenshots

National Technical University of Athens
-

University of Piraeus

25

Screenshots

National Technical University of Athens
-

University of Piraeus

26

Issues
-

Advantages

National Technical University of Athens
-

University of Piraeus

27


Platform unification creates super
-
layers of information,
abundance of classifications


Better and less costly knowledge management due to
knowledge representation


Exploitation of mobile communications rich in information, both
within the
company & the
outside world


More intelligent in
-
house HR monitoring


Rule
-
based reasoning for dynamically changing roles and
permissions


Rules can be modified, activated and deactivated on demand,
and rule notation frameworks (e.g. SPIN) make them easily
manageable


Easy corporate data exchange


Issues


Drawbacks

National Technical University of Athens
-

University of Piraeus

28


Costly development and maintenance of components for all platforms


Automatic capturing of threats from social media not mature enough,
need for manual monitoring


Semantic web services also not mature enough in e
-
commerce


Extraction of communications and external data is web
-
dependent


Need for deployment of centralized RDF store (difficult for mobile phones
to act as RDF stores/SPARQL endpoints etc.)

Future Work

National Technical University of Athens
-

University of Piraeus

29


Expand on the proposed components


Incorporate user
-
friendly and domain
-
specific methods for creating rule
sets


Create tools for assessment and analysis of
communication
patterns


Examine migration of ERP databases to RDF stores


For questions, code,
etc
:


m.meimaris@gmail.com

or

m.meimaris@medialab.ntua.com

National Technical University of Athens
-

University of Piraeus

30