Differences

This shows you the differences between two versions of the page.

--- teaching:mfe:is [2015/04/13 14:46]
svsummer [Engineering a runtime system and compiler for AQL]
+++ teaching:mfe:is [2018/04/30 11:06]
svsummer
@@ Line 1: / Line 1: @@
-====== MFE 2015-2016 : Web and Information Systems ======
+====== MFE 2018-2019 : Web and Information Systems ======
 ===== Introduction =====
@@ Line 18: / Line 18: @@
 ===== Master Thesis in Collaboration with Euranova =====
-Our laboratory performs collaborative research with Euranova R&D (http://euranova.eu/). The list of subjects proposed for this year by Euranova can be found
+Our laboratory performs collaborative research with Euranova R&D (http://euranova.eu/). The list of subjects proposed for this year by Euranova can be found {{:teaching:mfe:euranova_masterthesis_2017.pdf|here}}.
-{{:teaching:mfe:mt2014_euranova.pdf|here}}
 These subject include topics on distributed graph processing, processing big data using Map/Reduce, cloud computing, and social networks.
@@ Line 25: / Line 25: @@
   * Contact : [[ezimanyi@ulb.ac.be|Esteban Zimanyi]]
-===== Compiling SPARQL queries into machine code =====
+===== Dynamic Query Processing on GPU Accelerators =====
-Due to the increasing availability of larger and larger cheap RAM memories, the working set of modern database management systems becomes more and more main memory resident. This implies that, in contrast to traditional database management systems, slow disk accesses are rare, and that hence, the in-memory processing speed of databases becomes an important factor. As recently observed by a number of researchers, (e.g., [[http://sites.computer.org/debull/A14mar/p3.pdf|Neumann and Leis]]), one very attractive approach for fast query processing in this context is the just-in-time compilation of incoming queries into machine code. This compilation avoids the overhead of the traditional interpretation of query plans, and can aid in minimzing memory traffic for boosting performance.
+This master thesis is put forward in the context of the DFAQ Research Project: "Dyanmic Processing of Frequently Asked Queries", funded by the Wiener-Anspach foundation.
-A number of recent research prototypes exist that compile SQL queries into machine code in this sense:  HyPer A Hybrid OLTP&OLAP High Performance DBMS (http://hyper-db.de/) and Legobase (https://github.com/epfldata/NewLegoBase and http://data.epfl.ch/legobase).
+Within this project, our lab is hence developing novel ways for processing "fast Big Data", i.e., processing of analytical queries where the underlying data is constantly being updated. The analytics problems envisioned cover wide areas of computer science and include database aggregate queries, probabilistic inference, matrix chain computation, and building statistical models.
-The objective of this master thesis is to apply the same methodology to engineer a compiler that translates (fragments of) SPARQL (the standard query language for querying RDF data on the semantic web) into machine code. The overall methodology should follow the methodology used by HyPer and Legobase:
+The objective of this master thesis is to build upon the novel dynamic processing algorithms being developed in the lab, and complement these algorithms by proposing dynamic evaluation algorithms that execute on modern GPU architectures, thereby exploiting their massive parallel processing capabilities.
-  * Use of a high-level language to construct the compiler (Scala, http://scala-lang.org/)
-  * Use of Latent Modular Staging (LMS for short) for generating low-level portable assembly code at runtime (http://scala-lms.github.io/)
-  * Use of LLVM (http://llvm.org/) as a portable assembly code and corresponding translator to machine code.
-Getting aquaintend with these technologies is part of the master thesis objective.
+Since our current development is done in the Scala programming language, prospective students should either know Scala, or being willing to learn it within the context of the master thesis.
-**Validation of the approach** The thesis should propose a benchmark collection of SPARQL queries that can be used to test the obtained SPARQL-to-machine-code compiler and compare its perforance against a reference, interpreter-based SPARQL compiler.
+**Validation of the approach** Validation of master thesis' work should be done on two levels:
+  * a theoretical level; by proposing and discussing alternative ways to do incremental computation on GPU architectures, and comparing these from a theoretical complexity viewpoint
+  * an experimental level; by proposing a benchmark collection of CEP queries that can be used to test the obtained versions of the interpreter/compiler, and report on the experimentally observed performance on this benchmark.
-**Deliverables** of the master thesis project:
+**Deliverables** of the master thesis project
-  - An overview of the state of the art in query-to-machine-code compilation.
+  * An overview of query processing on GPUs
-  - A description of latent modular staging and how it can be used to construct machine-code compilers.
+  * A definition of the analytics queries under consideration
-  - The SPARQL compiler (software artifact)
+  * A description of different possible dynamic evaluation algorithms for the analytical queries on GPU architectures.
-  - A benchmark set of SPARQL queries and associated data sets for the experimental validation
+  * A theoretical comparison of these possibilities
-  - An experimental validation of the compiler, comparing efficiency of compiled queries against a reference compiler based on query plan interpretation.
+  * The implementaiton of the evaluation algorithm(s) (as an interpreter/compiler)
+  * A benchmark set of queries and associated data sets for the experimental validation
+  * An experimental validation of the compiler, and analysis of the results.
+**Interested?**
-**Interested?** Contact : [[stijn.vansummeren@ulb.ac.be|Stijn Vansummeren]]
+  * Contact : //Stijn Vansummeren//
 **Status**: available
-===== An implementation of the SCULPT schema language for tabular data on the Web =====
+===== Multi-query Optimization in Spark =====
-Despite the availability of numerous standardized formats for semi-structured and semantic web data such as XML, RDF, and JSON, a very large percentage of data and open data published on the web, remains tabular in nature. (Jeni Tennison, one of the two co-chairs of the W3C CSV on the Web working group claims that ``over 90% of the data published on data.gov.uk is tabular data''.) Tabular data is most commonly published in the form of comma separated values (CSV) files because such files are open and therefore processable by numerous tools, and tailored for all sizes of files ranging from a number of KBs to several TBs. Despite these advantages, working with CSV files is often cumbersome because they are typically not accompanied by a //schema// that describes the file's structure (i.e., ``the second column is of integer datatype'', ``columns are delimited by tabs'', etc) and captures its intended meaning. Such a description is nevertheless vital for any user trying to interpret the file and execute queries or make changes to it.
+Distributed computing platforms such as Hadoop and Spark focus on addressing the following challenges in large systems: (1) latency, (2) scalability, and (3) fault tolerance. Dedicating computing resources for each application executed by Spark can lead to a waste of resources. Unified distributed file systems such as Alluxio has provided a platform for computing results among simultaneously running applications. However, it is up to the developers to decide on what to share.
-In other data models, the presence of a schema is also important for query optimization (required for scalable query execution if the file is large), as well as other static analysis tasks. Finally, schemas are a prerequisite for unlocking huge amounts of tabular data to the Semantic Web.
+The objective of this master thesis is to optimize various applications running on a Spark platform, optimize their execution plans by autonomously finding sharing opportunities, namely finding the RDDs that can be shared among these applications, and computing these shared plans once instead of multiple times for each query.
-In recognition of this problem, the CSV on the Web Working Group of the World Wide Web Consortium argues for the introduction of a schema language for tabular data to ensure higher interoperability when working with datasets using the CSV or similar formats.
+**Deliverables** of the master thesis project
+  * An overview of the Apache Spark architecture.
+  * Develop a performance model for queries executed by Spark.
+  * An implementation that optimizes queries executed by Spark and identify sharing opportunities.
+  * An experimental validation of the developed system.
-The objective of this master thesis is to implement a recent proposal for such a schema language named SCULPT (http://arxiv.org/abs/1411.2351). Concretely, this entails:
+**Interested?**
-  * proposing an elegant concrete syntax for SCULPT schemas
+  * Contact : //Iman Elghandour// or //Stijn Vansummeren//
-  * implement both the in-memory and streaming validation algorithms of SCULPT proposed in http://arxiv.org/abs/1411.2351
-  * extend the SCULPT proposal, by investigating how SCULPT can be combined with complementary features recently proposed by the W3C CSV on the Web Working group (http://www.w3.org/2013/csvw/wiki/Main_Page)
-  * and in particular, extend sculpt with features that allow tabular files to be converted into RDF
-  * create associated tooling for SCULPT (i.e., parser and serializer generator, in the spirit of data description tools)
-\\
-**Deliverables** of this master thesis project:
-  - detailed description of the SCULPT proposal (document)
-  - overview of the state of the art; in particular other proposals for schema languages for tabular data (document)
-  - concrete syntax for sculpt (design document + formal grammar)
-  - implementation of SCULPT validation algorithms (software artifact)
-  - extension of sculpt with features for converting into RDF (document + software)
-**Interested?** Contact: [[stijn.vansummeren@ulb.ac.be|Stijn Vansummeren]]
 **Status**: available
-===== Engineering a runtime system and compiler for AQL =====
+===== Accelerated Distributed Platform for Spatial Queries =====
-Automatically extracting structured information from text is a task that has been pursued for decades.Since most analytics over text involves information extraction as a first step, IE is a very important part of data analysis in the enterprise today.
+It is now common to query terabytes of spatial data. Several new frameworks extend distributed computing platforms such as Hadoop and Spark to enable them to efficiently process spatial queries by providing (1) mechanisms to efficiently store spatial data and index them ; and (2) packages of built in spatial operations for these platforms. Meanwhile, it is now common to accelerate Hadoop and Spark using accelerators such as GPUs and FPGAs.
-In 2005, researchers at the IBM Almaden Research Center developped a new system specifically geared for practical information extraction in the enterprise. This effort lead to SystemT, a rule-based IE system with an SQL-like declarative language named AQL (Annotation Query Language). The declarative nature of AQL enables new kinds of tools for extractor development, and draws upon known techniques form query processing in relational database management systems to offer a cost-based optimizer that ensures high-througput performance. Recent research into the foundations of AQL (http://researcher.watson.ibm.com/researcher/files/us-fagin/jacm15.pdf) has shown that, as an alternative, it is also possible to build a runtime system for AQL based on special kinds of finite state automata. A potential benefit of this alternate runtime system is that text files need only be processed once (instead of multiple times in the cost-based optimizer backend) and may hence provide greater throughput. On the other hand, the alternate system can sometimes have larger memory requirements than the cost-based optimizer backend.
+The objective of this master thesis is to build a framework that efficiently executes spatial queries on a Spark version that is enabled to run its tasks on GPUs.
-The objective of this master thesis is to design and engineer a runtime system and compiler for (a fragment) of AQL based on finite state automata. Ideally, to obtain the best performance, these automata should be compiled into machine-code when executed. For this compilation, the following technologies should be used:
+**Deliverables** of the master thesis project
-  * A a high-level language to construct the compiler (Scala, http://scala-lang.org/)
+  * An overview of Spatial queries and frameworks for processing big spatial data.
-  * Use of Latent Modular Staging (LMS for short) for generating low-level portable assembly from the automata at runtime (http://scala-lms.github.io/)
+  * A study of best approaches to represent spatial data while it is queried by Spark and GPUs.
-  * Use of LLVM (http://llvm.org/) as a portable assembly code and corresponding translator to machine code.
+  * An implementation of common spatial operations and computational geometry algorithm on GPUs and Spark.
+  * An experimental validation of the developed system.
-Getting aquaintend with these technologies is part of the master thesis objective.
+**Interested?**
+  * Contact : //Iman Elghandour// or //Stijn Vansummeren//
-**Validation of the approach** The thesis should propose a benchmark collection of AQL queries and associated input text files that can be used to test the obtained automaton-based AQL compiler and compare its performance against the reference, cost-based optimizer of SystemT.
-**Deliverables** of the master thesis project:
-  - An overview of AQL, SystemT, and its cost-based optimizer and evaluation engine. (document)
-  - A description of how AQL can be evaluated by means of so-called vset finite state automata. (document)
-  - A detailed desription of the state of the art in evaluating finite state automata. (document)
-  - Identification of the AQL syntaxt that is to be supported. (specification)
-  - The AQL compiler (software artifact)
-  - A benchmark set of AQL queries and associated data sets for the experimental validation
-  - An experimental validation of the compiler, comparing efficiency of compiled queries against the cost-based reference compiler.
-\\
-**References about SystemT**:
-  * [[http://almaden.ibm.com/cs/projects/avatar/icde2008.pdf|An Algebraic Approach to Rule-Based Information Extraction]]
-  * [[http://www.sigmod.org/publications/sigmod-record/0812/p007.special.krishnamurthy.pdf|SystemT: A System for Declarative Information Extraction]]
-\\
-**References about finite state automata evaluation**:
-  * Regular expression pattern matching can be simple and fast. http://swtch.com/~rsc/regexp/regexp1.html
-  * Regular Expression Matching: the Virtual Machine Approach http://swtch.com/~rsc/regexp/regexp2.html
-  * Regular Expression Matching in the Wild http://swtch.com/~rsc/regexp/regexp3.html
-  * [[http://www.diku.dk/kmc/documents/AiPL-CrashCourse.pdf|A Crash-Course in Regular Expression Parsing and Regular Expressions as Types.]]
-\\
-**Interested?** Contact : [[stijn.vansummeren@ulb.ac.be|Stijn Vansummeren]]
-\\
 **Status**: available
+===== Co-locating Big Spatial Data Stored in HDFS =====
-===== Structural compression of relational databases =====
+Spatial databases employ spatial indexes to speedup the access of spatial data. New frameworks are introduced to build such indexes for Hadoop and Spark. However, there are not fully integrated on the file system level.
-Recent research in database management systems at ULB has shown how to theoretically construct succinct (compressed) representations for relational databases and semantic web databases. The advantage of these succinct representations is that they allow querying directly **on the succinct representation**, without needing to consult the underlying database.
+The objective of this master thesis is to build these indexes within the layer of HDFS and use this implementation to co-locate files that are typically accessed together by the spatial queries.
-The goal of this thesis is to study scalable algorithms for constructing the actual succinct representations. Some in-memory algorithms are already known, but given the large size of typical database, distributed and out-of-core alternatives need to be found.
+**Deliverables** of the master thesis project
+  * An overview of spatial queries and frameworks for processing big spatial data.
+  * A study of different types of indexes how they can be built in HDFS, and how we can use the replicas of HDFS to store multiple types of indexes
+  * An implementation of spatial indexes in HDFS.
+  * An experimental validation of the developed system.
-**Deliverables**:
+**Interested?**
-  * Overview of the state of the art in main-memory, and distributed (bi)simulation-based compression algorithms (document)
+  * Contact : //Iman Elghandour// or //Stijn Vansummeren//
-  * Description of the simulation-based compression algorithm to implement (document)
-  * Selection of the distribution framework (Actors, Pregel, ...) (document)
-  * Simulation algorithm (software artifact)
-  * Experimental analysis of distributed algorithm on a number of datasets. (document)
-**Interested?** Contact : [[stijn.vansummeren@ulb.ac.be|Stijn Vansummeren]]
 **Status**: available
-===== A Scala-based runtime and compiler for Distributed Datalog =====
+===== Complex Event Processing in Apache Spark and Apache Storm =====
-Datalog is a fundamental query language in datamanagement based on logic programming. It essentially extends select-from-where SQL queries with recursion. There is a recent trend in data management research to use datalog to specify distributed applications, most notably on the web, as well as do inference on the semantic web. The goal of this thesis is to engineer a basic **distributed datalog system**, i.e., a system that is capable of compiling & running distributed datalog queries. The system should be implemented in the Scala programming language. Learning Scala is part of the master thesis project.
+The master thesis is put forward in the context of the SPICES "Scalable Processing and mIning of Complex Events for Security-analytics" research project, funded by Innoviris.
-The system should incorporate recently proposed worst-case join algorithms (i.e., the [[http://arxiv.org/abs/1210.0481|leapfrog trie join]]) and employ known local datalog optimizations (such as magic sets and QSQ.)
+Within this project, our lab is developping a declarative language for Complex Event Processing (CEP for short). The goal in Complex Event Processing is to derive pre-defined patterns in a stream of raw events. Raw events are typically sensor readings (such as "password incorrect for user X trying to log in on machine Y" or "file transfer from machine X to machine Y"). The goal of CEP is then to correlate these events into complex events. For example, repeated failed login attempts by X to Y should trigger a complex event "password cracking warning" that refers to all failed login attempts.
-**Validation of the approach** The thesis should propose a benchmark collection of datalog queries and associated data workloads that be used to test the obtained system, and measure key performance characteristics (elasticity of the system; memory frootprint; overall running time, ...)
+The objective of this master thesis is to build an interpreter/compiler for this declarative CEP language that targets the distributed computing frameworks Apache Spark and/or Apache Storm as backends. Getting aquaintend with these technologies is part of the master thesis objective.
+**Validation of the approach** Validation of the proposed interpreter/compiler should be done on two levels:
+  * a theoretical level; by comparing the generated Spark/Storm processors to a processor based on "Incremental computation" that is being developped at the lab
+  * an experimental level; by proposing a benchmark collection of CEP queries that can be used to test the obtained interpreter/compiler, and report on the experimentally observed performance on this benchmark.
-**Deliverables**:
+**Deliverables** of the master thesis project
-  * Semantics of datalog; overview of known optimization strategies (document)
+  * An overview of the processing models of Spark and Storm
-  * Description of the leapfrog trie join (document)
+  * A definition of the declarative CEP language under consideration
-  * Datalog system (software artifact)
+  * A description of the interpretation/compilation algorithm
-  * Experimental analysis of developped system on a number of use cases (document)
+  * A theoretical comparison of this algorithm wrt an incremental evaluation algorithm.
+  * The interpreter/compiler itself (software artifact)
+  * A benchmark set of CEP queries and associated data sets for the experimental validation
+  * An experimental validation of the compiler, and analysis of the results.
-\\
+**Interested?**
-**Interested?** Contact : [[stijn.vansummeren@ulb.ac.be|Stijn Vansummeren]]
+  * Contact : //Stijn Vansummeren//
 **Status**: available
-===== Design and Implementation of a Curriculum Revision Tool =====
+===== Complex Event Processing in Apache Spark and Apache Storm =====
-Stijn Vansummeren (WIT), Frédéric Robert (BEAMS)
+The master thesis is put forward in the context of the SPICES "Scalable Processing and mIning of Complex Events for Security-analytics" research project, funded by Innoviris.
-This master thesis project concerns the analysis, design, and implementation of a software system that can assist in the revision of teaching curricula (also known as teaching programs).
+Within this project, our lab is developping a declarative language for Complex Event Processing (CEP for short). The goal in Complex Event Processing is to derive pre-defined patterns in a stream of raw events. Raw events are typically sensor readings (such as "password incorrect for user X trying to log in on machine Y" or "file transfer from machine X to machine Y"). The goal of CEP is then to correlate these events into complex events. For example, repeated failed login attempts by X to Y should trigger a complex event "password cracking warning" that refers to all failed login attempts.
-The primary targetted functionalities of the software system are as follows:
+The objective of this master thesis is to build an interpreter/compiler for this declarative CEP language that targets the distributed computing frameworks Apache Spark and/or Apache Storm as backends. Getting aquaintend with these technologies is part of the master thesis objective.
-  * It should allow to make different versions of the teaching programs, much in the same way as version control systems like GIT and subversion offer the possibility to make different "development branches" of a program's source code.
+**Validation of the approach** Validation of the proposed interpreter/compiler should be done on two levels:
-  * It should allow an extensible means to check the modified program for inconsistentcies. (For example, if course X has course Y as prerequisite, then course Y should not be scheduled in 2nd semester and X in 1st semester. Moreover, the total number of ECTS of all courses should be at most 60 ECTS. )
+  * a theoretical level; by comparing the generated Spark/Storm processors to a processor based on "Incremental computation" that is being developped at the lab
-  * It should allow to analyze the modifications proposed in the teaching programs, and summarize the impact that these changes could have on other programs. (For example, if a course is removed from the computer science curriculum, it should be flagged that it should also be removed from all curricula that included the course.)
+  * an experimental level; by proposing a benchmark collection of CEP queries that can be used to test the obtained interpreter/compiler, and report on the experimentally observed performance on this benchmark.
-  * It should load data from (and preferably, save data to) the ULB central administration database.
-  * It should give suggestions concerning the impact of the modifications on the course schedules.
-A proof-of-concept implementation of a revision tool that supports the first two requirements above is currently being developped in the context of a PROJH402 project. The MFE student that selects this topic is expected to:
+**Deliverables** of the master thesis project
-  * Develop this prototype to a production-ready implementation.
+  * An overview of the processing models of Spark and Storm
-  * Implement the communication with the central ULB database.
+  * A definition of the declarative CEP language under consideration
-  * Implement the impact analysis concerning the course schedules.
+  * A description of the interpretation/compilation algorithm
-  * Interact with the administration of the Ecole Polytechnique to fine-tune the above requirements; test the implementation; and integrate remarks after testing
+  * A theoretical comparison of this algorithm wrt an incremental evaluation algorithm.
+  * The interpreter/compiler itself (software artifact)
+  * A benchmark set of CEP queries and associated data sets for the experimental validation
+  * An experimental validation of the compiler, and analysis of the results.
-\\
+**Interested?**
-**Interested?** Contact : Stijn Vansummeren (stijn.vansummeren@ulb.ac.be), Frédéric Robert <frrobert@ulb.ac.be>
+  * Contact : [[stijn.vansummeren@ulb.ac.be|Stijn Vansummeren]]
+**Status**: available
-===== Automatic detection of name variations =====
-Toon Calders (WIT)
-For this project a large data collection consisting of historical birth, death, and marriage certificates of the province of North-Brabant in the Netherlands is available. This collection contains certificates for about 3 million people, from 1580 until 1955. This collection of paper documents has been indexed by volunteers. For many of the certificates (unfortunately the index is not complete yet), the names of the people involved in it, and their role have been recorded in a database. Consider for instance the following example of an index entry for a death certificate:
+===== Graph Indexing for Fast Subgraph Isomorphism Testing =====
-^ Death certificate ^^
+There is an increasing amount of scientific data, mostly from the bio-medical sciences, that can be represented as collections of graphs (chemical molecules, gene interaction networks, ...). A crucial operation when searching in this data is that of subgraph    isomorphism testing: given a pattern P that one is interested in (also a graph) in and a collection D of graphs (e.g., chemical molecules), find all graphs in G that have P as a   subgraph. Unfortunately, the subgraph isomorphism problem is computationally intractable. In ongoing research, to enable tractable processing of this problem, we aim to reduce the number of candidate graphs in D to which a subgraph isomorphism test needs   to be executed. Specifically, we index the graphs in the collection D by means of decomposing them into graphs for which subgraph   isomorphism *is* tractable. An associated algorithm that filters graphs that certainly cannot match P can then formulated based on ideas from information retrieval.
-|Deceased	|Johanna Louise Fredrika Frans |
-|Relation of the deceased |Gerard Cornelius Reincke de Sitter |
-|Father of the deceased	|Carl Ludwig Frans |
-|Mother of the deceased	|Alida Philippina Zehender |
-|Type of deed |death certificate |
-|Number of deed	|5 |
-|Place	|Beers |
-|Date of decease |26-02-1825 |
-|Period	|1825 |
-|Contains |Overlijdensregister 1825 |
-|Number of inventory |50 |
-|Record number	|456 |
-There are, however, several problems with the data recorded by the volunteers:
+In this master thesis project, the student will emperically validate on real-world datasets the extent to which graphs can be decomposed into graphs for which subgraph isomorphism is tractable, and run experiments to validate the effectiveness of the proposed method in terms of filtering power.
-  - Volunteers made mistakes when recording the names
-  - Natural name variations occur; for instance, during the Napoleonic era, Willem preferred to be called Guillaume. After the French left the Netherlands, Willem became Willem again. Other, less spectacular variations: Fredrika versus Frederika.
-  - Another source of variation is the granularity at which locations are reported. Sometimes locations have been reported at suburb or even neighborhood level, whereas in other records only the city is reported.
-  - Also the original data contained errors. For instance, the order of names may have been swapped.
-The goal of this graduation project is to automatically detect name variations for location and person names, using statistical and data mining methods. Because of the large size of the database it is very likely that most name variations occur frequently. In a pilot study, it was shown that name variations could be detected by finding pairs of full names sharing most surnames, but not all. The differences often were name variations. Your task will be to extend this approach to also include locations, and exploit additional background knowledge such as: for most birth certificates there is a matching death certificate, no one has more than one birth and death certificate, etc.
+**Interested?** Contact : [[stijn.vansummeren@ulb.ac.be|Stijn Vansummeren]]
-This project has a large research component, so your creative input will be required as well. For this project it is absolutely not necessary to speak or understand Dutch.
-Interested? Contact [[toon.calders@ulb.ac.be|Toon Calders]]
+**Status**: available
-===== Analyzing state-of-the-art technology for handwritten text recognition in a practical case study =====
-Toon Calders (WIT) and Olivier Debeir (LISA)
-The goal of this project is to study the applicability of current state-of-the-art text recognition tools in the following practical application. Consider the following two exemplary documents:
+=====Sentiment Analysis=====
-[[https://dl.dropbox.com/u/5119252/MFE/069-50-3165-1813-00009.jpg]] \\
-[[https://dl.dropbox.com/u/5119252/MFE/069-50-3165-1815-00003.jpg]]
-These two documents are scans of birth certificates (actually both are 2 birth certificates) from the Dutch city Grave. We have a huge collection of such paper documents; about 3 million, of which several tens of thousands have been scanned. Furthermore, we have an index on these documents, created by volunteers. This index contains, for the birth certificate, the name of the child, the name of the father and mother, and the witnesses. As you can see in the documents, however, much more information is available. Your task is to answer the following question: is it realistic, given the current state-of-the-art to do automatic recognition of hand-written texts such as these certificates? Most of the documents are very structured, with limited number of possible values (age of a person, profession), and there is a huge amount of training data; the names of all people have been indexed, usually the handwriting is consistent throughout a whole book with certificates. This graduation project includes a thorough literature study and experimentation with (original combinations of) state-of-the-art image recognition techniques adapted to our specific case. The project will be carried out in collaboration with the research labs WIT and LISA.
+The sentiment analysis task aims to detect subjective information polarity in the target text by applying Natural Language Processing (NLP), text analysis and computational linguistics techniques. With the emergence of web 2.0, it becomes easy for Internet users to post their opinionated comments and share their thoughts via social networks, forums and especially Twitter. With more resources and NLP tools becoming available and with the recent developed sentiment lexicons, sentiment analysis is having more attention from the research community. Nevertheless, Named Entities (NEs) effectiveness was not studied even though it is easily noticeable that social resources include many NEs. In ongoing research, we aim to investigate the effectiveness of Named Entities (person, location and organization entities) on sentiment analysis and dive beyond the Named Entities recognition to propose a framework of Named Entities polarity classification and process an empirical evaluation on their effectiveness on Sentiment classification.
-Interested? Contact [[toon.calders@ulb.ac.be|Toon Calders]]
+In this master thesis project, the student will empirically validate on real-world datasets the effectiveness of Named Entities (person, location and organization entities) on sentiment analysis and run experiments on different languages (French, Dutch, English and German).
-===== Process Mining on Company Data for Detecting Security Breaches =====
+**Interested?** Contact : [[haddad.hatem@gmail.com|Hatem Haddad]]
-Toon Calders (WIT)
-According to a recent report of Price Waterhouse Cooper, the most common source of security incidents are current employees, followed at a distance by former employees and only after that truly external threats such as hactivists. [http://www.pwc.com/gx/en/consulting-services/information-security-survey/giss.jhtml?region=&industry=]  This observation leads to the conclusion that in an intelligent security event management system, should also concentrate on internal threats to security.
-The goal of the thesis is to analyze the possibility of using process mining to help in the detection of silent attacks. We will concentrate on company-specific data. From this data typical behavior will be detected and modeled as a process or workflow. We consider three aspects of a workflow: the actor(s), the resources, and the activities. By modeling the normal behavior in the system we are able to detect deviating cases. Based on historical data, the goal is to build models of typical behavior, including the use of resources such as patient records. Such a system would be able to detect for instance if a certain patient record is consulted much more often than usual, or by more people, or outside of the normal workflow (e.g., only reading information, but not writing). Such a pattern could indicate unjustified access to for instance the patient record of a famous patient.
-For modeling the workflows, we propose the use of process mining (Van der Aalst, 2011). Process mining is a state-of-the-art technology concerned with the automatic extraction of process models from event logs. Consider, e.g., a hospital registering all activities that are carried out for the treatment of patients, ranging from the admission, various measurements being taken from the patient, medicine administered, surgical procedures, to the resignation of the patient. Process mining could be used to extrapolate from these examples, a common model of how the hospital deals with a patient. There are several applications of process mining; first it can be used to improve the processes by standardizing them; many companies and organizations may only have informal procedures. By process mining the process logs are used to extract a general model of the actual business processes. Such a model can guide the automation process.
-In this thesis the goal is to analyze how process mining could be used for anomaly detection; how can the discovered models be used to detect abnormal behavior in a company network? Much like in credit card fraud detection, the approach is to first model normal behavior, in this case using process mining, in order to detect diverging behavior that could indicate security breaches in the network.
-Van der Aalst, W. M. (2011). Process Mining: Discovery, Conformance and Enhancement of Business Processes. Springer.
-Interested? Contact [[toon.calders@ulb.ac.be|Toon Calders]]
-===== Mining patterns for compression =====
-Toon Calders (WIT)
-Data mining is the research discipline that studies the extraction of information from large amounts of data. One of the typical data mining tasks is pattern mining where we try to find regularities that occur frequently in a dataset. The prototypical example is that of a supermarket storing for every customer visiting the supermarket, the transaction; that is, the set of items that were bought by that customer. The frequent itemset mining problem now is to detect which combinations of products were more often sold together than a given threshold. One of the major problems of pattern mining algorithms, however, is the enormous amount of redundant patterns they generate; for instance, very popular items, such as toilet paper, tend to appear in many frequent combinations purely due to chance. In order to deal with this problem, techniques based upon compression and minimum description length were proposed to reduce the number of patterns. The rationale behind the minimal description length principle is that a set of patterns that describes well what is happening in the dataset should allow for a good compression. For a collection of patterns, the quality is measured as the description length of the patterns plus the size of the data compressed with these patterns. For instance, if the pattern {bread, milk, butter} has a high frequency, we could opt to replace every occurrence of this pattern by a special code, effectively reducing the encoding length of the data. Surprisingly, however, the MDL principle was until now only used to rule out redundant patterns, and it has not been researched yet how well the discovered patterns actually do compress the data as compared to compression algorithms such as Lempel–Ziv–Welch.
-Hence, in this highly research oriented graduation project, two research questions are central: (1) How good do non-redundant pattern sets based on MDL allow compressing data, and (2) Can we extract useful patterns from existing compression algorithms?
-Interested? Contact [[toon.calders@ulb.ac.be|Toon Calders]]
-===== Pattern Mining for Object Tracking =====
-Toon Calders (WIT)
-Pattern mining techniques are more and more often used in computer vision
-to obtain features that are more discriminative than those extracted
-using computer vision algorithms. This is true for example in content-based
-images/videos retrieval, indexing, classiﬁcation, tracking, etc. However, the main
-drawback of using traditional pattern mining techniques is their ineﬃciency when
-dealing with huge set of data (for example provided by Google image or Youtube
-for videos) or when trying to tackle real-time analysis problems. The data mining
-community has been working on the “Big Data” problem for many years coming
-up with promising solutions such as stream mining. The aim of this project
-is to explore the possibility of using pattern mining in data streams for the (real-time) analysis of videos and, in particular, for object tracking.
-For more extensive information regarding the context and problem setting, see the following paper:
-Toon Calders, Elisa Fromont, Baptiste Jeudy and Hoang Thanh Lam.
-[[http://labh-curien.univ-st-etienne.fr/~fromont/|Analysis of Videos using Tile Mining.]]\\
-In: //ECML/PKDD Workshop on Real-World Challenges for Data Stream Mining//, Prague, 2013
-Interested? Contact [[toon.calders@ulb.ac.be|Toon Calders]]
-===== Design and Implementation of a Curriculum Revision Tool ======
-Stijn Vansummeren (WIT), Frédéric Robert (BEAMS)
-This MFE concers the analysis, design, and implementation of a
-software system that can assist in the revision of teaching curricula
-(also known as teaching programs).
-The primary targetted functionalities of the  software system are as
-follows:
-  * It should allow to make different versions of the teaching programs, much in the same way as version control systems like GIT and subversion offer the possibility to make different "development branches" of a program's source code.
-  * It should  allow an extensible means to check the modified program for inconsistentcies. (For example, if course X has course Y as prerequisite, then course Y should not be scheduled in 2nd semester and X in 1st semester. Moreover, the total number of ECTS of all courses should be at most 60 ECTS. )
-  * It should allow to analyze the modifications proposed in the teaching programs, and summarize the impact that these changes could have on other programs. (For example, if a course is removed from the computer science curriculum, it should be flagged that it should also be removed from all curricula that included the course.)
-  * It should load data from (and preferably, save data to) the ULB central administration database.
-  * It should give suggestions concerning the impact of the modifications on the course schedules.
-A proof-of-concept implementation of a revision tool that supports the first two requirements above is currently being developed in the context of a PROJH402 project. The MFE student that selects this topic is expected to:
-  * Develop this prototype to a production-ready implementation.
-  * Implement the communication with the central ULB database.
-  * Implement the impact analysis concerning the course schedules.
-  * Interact with the administration of the Ecole Polytechnique to fine-tune the above requirements; test the implementation; and integrate remarks after testing
-Contact : Stijn Vansummeren <stijn.vansummeren@ulb.ac.be>, Frédéric Robert <frrobert@ulb.ac.be>
+**Status**: available
 =====Publishing and Using Spatio-temporal Data on the Semantic Web=====
@@ Line 318: / Line 201: @@
    * Contact: [[ezimanyi@ulb.ac.be|Esteban Zimányi]]
+=====Efficient Management of (Sub-)structure  Similarity Search Over Large Graph Databases. =====
+The problem of (sub-)structure similarity search over graph data has recently drawn significant research interest due to its importance in many application areas such as in Bio-informatics, Chem-informatics, Social Network, Software Engineering, World Wide Web, Pattern Recognition, etc.  Consider, for example, the area of drug design, efficient techniques are required to query and analyze huge data sets of chemical molecules thus shortening the discovery cycle in drug design and other scientific activities.
+Graph edit distance is widely accepted as a similarity measure of labeled graphs due to its ability to cope with any kind of graph structures and labeling schemes.  Today, graph edit similarity plays a significant role in managing graph data , and is employed in a variety of analysis tasks such as graph classification and clustering, object recognition in computer vision, etc.
+In this master thesis project,  due to the hardness of graph edit distance (computing graph edit distance is known to be NP-hard problem), the student  will investigate the current approaches that deals with problem complexity while searching for similar (sub-)structures.  At the end, the student should be able to empirically analyze and contrast some of the interesting approaches.
+=====A Generic Similarity Measure For Symbolic Trajectories=====
+Moving object databases (MOD) are database systems that can store and manage moving object data. A moving object is a value that changes over time. It can be spatial (e.g., a car driving on the road network), or non-spatial (e.g., the temperature in Brussels). Using a variety of sensors, the changing values of moving objects can be recorded in digital formats. A MOD, then, helps storing and querying such data. There are two types of MOD. The first is the trajectory database, that manages the history of movement. The second type, in contrast, manages the stream of current movement and the prediction of the near future. This thesis belongs to the first type (trajectory databases). The research in this area mainly goes around proposing data persistency models and query operations for trajectory data.
+A sub-topic of MOD is the study of semantic trajectories. It is motivated by the fact that the semantic of the movement is lost during the observation process. You GPS logger, for instance, would record a sequence of (lon, lat, time) that describe your trajectory. It won't, however, store the purpose of your trip (work, leisure, …), the transportation mode (car, bus, on foot, …), and other semantics of your trip. Research works have accordingly emerged to extract semantics from the trajectory raw data, and to provide database persistency to semantic trajectories.
+Recently, Ralf Güting et al. published a model called “symbolic trajectories”, which can be viewed as a representation of semantic trajectories:
+Ralf Hartmut Güting, Fabio Valdés, and Maria Luisa Damiani. 2015. Symbolic Trajectories. ACM Trans. Spatial Algorithms Syst. 1, 2, Article 7 (July 2015), 51 pages.
+A symbolic trajectory is a very simple structure composed of a sequence of pairs (time interval, label). So, it is a time dependent label, where every label can tell something about the semantics of the moving object during its associated time interval. We think this model is promising because of its simplicity and genericness.
+The goal of this thesis is to implement a similarity operator for symbolic trajectories. There are three dimensions of similarity in symbolic trajectories: temporal similarity, value similarity, and semantic similarity. Such an operator should be flexible to express arbitrary combinations of them. It should accept a pair of semantic trajectories and return a numerical value that can be used for clustering or ranking objects based on their similarity. Symbolic trajectories are similar to time series, except that labels are annotated by time intervals, rather than time points. We think that the techniques of time series similarity can be adopted for symbolic trajectories. This thesis should assess that, and implement a similarity measure based on time series similarity. The implementation is required to be done as an extension to PostGIS. We have already implemented some temporal types and operations on top of PostGIS, where you can start from.
+**Deliverables** of the master thesis project
+  * Reporting on the state of art of semantic trajectory similarity measures.
+  * Reporting on the state of art in time series similarity measures.
+  * Assessing the application of time series similarity to symbolic trajectories.
+  * Implementing symbolic trajectories on top of PostGIS.
+  * Implementation and evaluating the proposed symbolic trajectory similarity operator.
+**Interested?**
+  * Contact : [[ezimanyi@ulb.ac.be|Esteban Zimanyi]]
+**Status**: available