vldb09-BelieveAnn-Suciu

"Believe It or Not: Adding belief Annotations to Databases", W. Gatterbauer, M.Balazinska, N.Khoussainova, D.Suciu

• Define a belief database as a set of belief statements. It contains base information in the form of ground tuples, annotated with belief statements
  
• Describe a model of database annotations that allow users to annotate both the content and other users' annotations with belief
  
• Use of a concrete semantics to annotations that helps users engage in a structured discussion on content and each other's annotations.
  
• Assume, by default, that a user believes every belief statement that is in the database, unless stated otherwise
• The annotations are of the boolean logical form: "Bob believes that Alice saw X"; "Bob does not believe that Alice saw X"
  
• The experiment proved that the actual overhead of belief annotations can be significantly lower than their theoretic bound
  

Sigmod09-GrammarBasedDataCleaning-Arasu_kaushik

"A Grammar-based Entity Representation Framework for Data Cleaning", Arvind Arasu, Raghav Kaushik

• the framework presented is a programmable module that can be used to transform "dirty" input records to one or more "clean" output records with consistent representation of entities and sub-entities
• uses grammar based rules to decide which records are the same given the first name, last name and affiliation, knowing that all these information can be written in different ways although represent the same record.
  

ACM02DataQAssesment -Pipino

"Data Quality Assessment", L.Pipino, Y.Lee, R.Wang

Completeness can be defined as:

• schema completeness : entities and attributes are not missing from the schema
• column completeness : a function of missing values in a column of a table
• population completeness : given the set of values a column should have, count how many are missing
  

All can be measured as a ratio =1- #incomplet items/total #of items

WWW03qDriven-Web-Zeng

"Quality driven web services composition" , L. Zeng, B. Benatallah, M. Dumas, J. Kalagnanam, Q. Sheng, WWW Conference, 2003

• Quality criteria used: price, duration, reputation, reliability, availability
• Composite service =an aggregation of all the component services
• Composite service is modeled as a statechart with initial state, final state and paths to travel from the init-state to fin-state
• price =the amount of money that a service requester has to pay for executing the operation (come from the web service providers)
• duration =expected delay between the moment when a request is sent and the moment when the result are received; is the sum of the processing time and the transmission time
  
• reliability =the probability that a request is correctly responded within a maximum expected time frame (computed from historical data)
• availability =the probability that the service is accessible ( assumption: web services send notifications to the system about their running states )
• reputation: the average ranking given to the service by the end users
• MAIN IDEA: algorithms to select the optimal execution plans in terms of quality based on linear programing
• an experiment was conducted using synthetic data.
  

IQ2000infoQ-Naumann

"Assesment Methods for Information Quality Criteria"
Felix Naumann, Claudia Rolker -In Proceedings of the International Conference on Information Quality, 2000

• Previous work: IQ scores rely on questionnaires; IQ based on soudness and completeness of information sources. Algorithm provided to compute the IQ score but is still based on user inout to decide weather some information is correct or not.
• 3 IQ classes: the user, the source and the query process
• 3 assessment-oriented IQ criteria classes: subject-criteria scores, process-criteria scores, Object-criteria scores
  

COMMENTS:

• no experiments are provided
• User input can be seen as how many time someone accessed a particular source and how long did they stayed on that source (if this information can be extracted) or how many times user returned to that source