语义集成笔记

 

前段时间的笔记，没有太多自己的想法。欢迎交流。

2007-01-18

semantic integration:a survey of ontology-based approaches 2004

ontology: is some formal description of a domain of discourse,intended for sharing among different applications and expressed in a language

that can be used for resoning.

 

three dismensions of smantic-integration research:

1 mapping discovery

        genaeral upper ontology               SUMO              DOLCE

        heusitics-based or machine learing teniques

                concept names and natural-language descriptions

                class hierarchy

                property definitions(domains ranges restrictions)

                instances of classes

                class descriptions(as in DL-based toos)

2 declarative formal representings of mappings

        representing mapppings as instances in an ongology of mapppings

        defining bridging axioms in first-order logic to represent transformations

        using views to describe mappings from global ontology to local ontologies

3reasoning of mappings       goal of mappings

summary:semantic integration is similary to the issues that database and information-integration researches have been addressing.

现在的趋势就是把mapping做到编辑器里面，同时编辑多个本体，在他们之间创立mapping，支持manual, semi/automatic的方式去管理mapping。纯粹用机器也是不现实的。小量的可以手工，大量的如何自动？

neon和cbio项目就是这个趋势，我比较相信这两个project的趋势也是以后ontology发展方向的趋势

2007-01-19

a survey of schema-based matching approaches

most mappings between:db schema ,xml schema and catalogs on the web,ontology

 

schema mappings without explicitly semantics while ontology obeys some formal semantics

so in schema matching,we should guess the meaning encoded in the schemas.

they both provide a vocabulary of terms that describes a domain of interest and the both constrain the meaning of terms used in  the vocabulary .

 

semantic is used for structure parsing,to find relations

alignment: mapping set

future:

1 novel matching approaches which exploit schema-level information

2 instance-based approaches

3 evaluation benchmarks

 

 

2007-1-22

Introduction to the special issue on semantic integration

anhai doan ,alon halevy

in this paper,author refers to many papers in this field and gives each one sentance introduction.

current challenge:

1scalability with the size of schema

2user interaction: a schema matching system will never be completely autonomous          how to small the interation?

3mapping maintenance : schama change frequently

 

issue fall into four categories

1 shcama matching techniques

2 current technical challenges      :scalability maintenance

3 practice consideration            :system

4 the role of ontology

 

 

 

 

 

2007.1.24 Data Integration:The Teenage Years

Alon Halevy 's paper "Query Heterogeneous Information sources using Source Descriptions1996" was awarded VLDB 10-years Best Paper award.

This paper is his presentation in the conference 2006.

 

有113篇参考，可以找到相关领域的文章

未提及semantic ontology的问题，主要从数据角度介绍

 

The information manifold: provide a uniform query interface to a multitude of data sources.

Propoesed the method LAV. An information source is described as a view expression over the mediated schema.

The mediated schema is defined by Source Descriptions(related to DL ..)

 

Direction:

1 generating schema mappings

        AI-complete problem

      obtain information from schema

      machine learing: mapping may repeat

      instance level

2 adaptive query processing

        query optimization

3XML

        the nested aspect of xml

        semi-structured

4 model management

        provide an algebra for manipulating shcemas and mappings

5p2p data management

        no global mediated shcam

        distributed mechanism for sharing data

6AI

        descritpion logic offered more flexible mechanisms for representing a mediated schama and for semantic

        query optimization needed in such systems.

        AI planning

        machine learning

 

Industry:

EII : mediated schema

efficient query processing and integration for xml was in its infancy

 

Future:

our goal should be to create tools that facilitate data integration in a variety of scenarios. (不可能完全自动化)

 

dataspace:按照需要一步步整理数据，减少使用代价

uncertainty and lineage  数据的来龙去脉

reusing human attention  人的参与

 

 

数据库管理一个问题：流数据 ，即实时数据

 

 

2007-1-29 Ontology-Based integration of information-a survey of existing approaches

 

1 Role of the ontology

        content explication

                single ontology approach

                multiple ontologies

                hybrid approaches

        query model

        verification

 

2 Ontology representations

        variants of description logics   ,extension of DL including rules

        classical framebased representation languages

3 use of mappings

        connnection to information sources

                structure resemblance

                definition of terms

                structure enrichment

                meta-annotation

        inter-ontology mapping

                defined mappings

                lexical relations

                top-level grounding

                semantics correspondences

4 ontology engineering

        development methodology

        supporting tools

        ontology evolution

 

open questions:

1 mapping problem

2 methodologies