Figure 3- - uploaded by Katarzyna Maria Wegrzyn-Wolska
Content may be subject to copyright.
Source publication
The work presented in this paper concerns automatic information retrieval and extraction in precise field of electronics based on linguistic knowledge. The extraction and the filtering of data is carried out automatically by methods based on the construction of local grammar. To carry out an automatic search for the events in the corpus, a linguist...
Similar publications
Triple Graph Grammars (TGGs) are a technique for defining the correspondence between two different types of models in a declara-tive way. The power of TGGs comes from the fact that the relation between the two models cannot only be defined, but the definition can be made operational so that one model can be transformed into the other in either dire...
A new interactive evolutionary 3D design system is pre-sented. The representation is based on graph grammars, a fascinating and powerful formalism in which sub-graphs, nodes and edges are itera-tively rewritten by rules analogous to those of context-free grammars and shape grammars. The nodes of the resulting derived graph are labelled with Euclide...
This paper shows how the correspondence between a unordered dependency tree and a sentence that expresses it can be achieved by transforming the tree into a string where each linear precedence link corresponds to one specific syntactic relation. We propose a formal grammar with a distributed architecture that can be used for both synthesis and anal...
Acquisition of new terminology from specific domains and its adequate description within terminological dictionaries is a complex task, especially for languages that are morphologically complex such as Serbian. In this paper we present an approach to solving this task semi-automatically on basis of lexical resources and local grammars developed for...
Citations
... Dictionaries associate the word with a lemma and a series of grammar, semantic and inflexion codes. Generally grammar represents sequences of words and produces linguistic information, for example information on the syntactic structure [19]. The corpora of clients' forum are represented by automata, in which each state corresponds to a lexical analysis. ...
... Dictionaries associate the word with a lemma and a series of grammar, semantic and inflexion codes. Generally grammar represents sequences of words and produces linguistic information, for example information on the syntactic structure [19]. The corpora of clients' forum are represented by automata, in which each state corresponds to a lexical analysis. ...
... @BULLET Integration Frameworks integration of TM (ex. The Med- Scan [10], [19] system combines lexicons with syntactic and semantic templates in a general-purpose TM system to extract relationships between biomedical entities). Our work focus on three branches of biomedicine TM, Text Classification (TC), Named Entity Recognition (NER) and Synonym and Abbreviation Extraction. ...
The quantity of medical publications in the media such a tweets, blogs, and another content in social media is growing at an exponential rate. Exploring and analyzing this content has become a necessity. The objective of this paper is to discuss the variety of issues and challenges surrounding the perspectives regarding the use of Social Network Analyses and Text Mining methods for applications in E health and medecine. The article will first look at the directions taken in Social Media and Text Mining for medical science. First we present a review of the literature in Social Media and Text Mining analysis for medical purposes and then describe the work done for our prediction system for collecting and manipulating and Twitter data.
... @BULLET Integration Frameworks integration of TM (ex. The MedScan [9], [16] system combines lexicons with syntactic and semantic templates in a general-purpose TM system to extract relationships between biomedical entities). Our work focus on three branches of biomedicine TM, Text Classification (TC), Named Entity Recognition (NER) and Synonym and Abbreviation Extraction. ...
The quantity of biomedical publications is growing at an exponential rate. With such explosive growth of the content, it is more and more difficult to locate, retrieve and manage the resulting information. This is why text mining has become a necessity. The main goal of biomedical research is to put knowledge to practical use in the form of diagnoses, prevention, and treatment. It is important to pool the resources between the different individuals researching results. The objective of this paper is to discuss the variety of issues and challenges surrounding the perspectives regarding the use of Information Retrieval and Text Mining methods in biomedicine. The article will first look at the directions in biomedical Text Mining and then describe the work done for the BIAM project, the French on-line Medical Data Base.
... Finally we obtained three marks for each review, and those marks were not always the same. We used another classifier which correlated the three marks in order to obtain the final mark [5], [6], [7]. The final classifier only used the three marks so as not to repeat the characteristics which are used in previous classifications. ...
... Finally we obtain a logico-conceptual representation of the text [2], [10], [1]. Semantico-conceptual structures can be more or less broad, rich and complex and more or less am- biguous [5]. This part of the system was developed with Unitex application , the example of linguistic resources used is shown infigure 3. We use a linguistic analyzer Unitex to pre-treat, to lemmatize the words, to add synonyms, to detect negation, to add semantic classes to the words and lastly to build complex local grammars. ...
This paper describes the functions of a system designed for the assessment of movie reviews. Such a system enables the automatic collection, evaluation and rating of film crit- ics' opinions of movies. First the system searches and re- trieves probable movie reviews from the Internet, especially those expressed by prolific reviewers. Subsequently the sys- tem carries out an evaluation and rating of those movie re- views. Finally the system automatically associates a numer- ical mark to each review, this is the objective of the system. This data constitutes the input to the cognitive engine. Our system uses three different methods for classifying opinions in critics' reviews. We introduce two new methods based on linguistic knowledge. Results are then compared with the overall statistical method using Bays classifier. The last step is to combine the results obtained in order to make the final assessment as accurately as possible. With the growth of the Web, e-commerce has become very popular. A lot of websites offer on line sales and pro- pose object ratings to their clients, for films for example. People like to check out other users' recommendations be- fore making up their minds. Those profiles are very use- ful for the customers. The Recommender System was cre- ated (RS) in order to predict the potential choice of clients. RS allows people to make choices without any personal knowledge of the alternatives. Algorithms for suggestion are based on the experience and the opinion of other users. It is helpful to find recommendations from people who are familiar with the same problems, who have made similar choices in the past, whose perspective we value, or who are recognized experts (15). RS provides correspondences between the users who have similar profiles. A new user has to create their own profile. The RS will suggest a new limited choice based on the similar taste of other users. The results of RS must not be tampered with for commercial reasons because this would make people distrustful. The effectness of this sys- tem depends on the data's quality and quantity. Our system supplies user profiles which are necessary for the algorithms of the cognitive engine. The main goal of the developed system is to collect a huge base of film reviews and auto- matically attribute marks which express the sentiments of the writer. Each review receivs a new mark and a user pro- file. The result of this treatment is the creation of a user profile database. Our system is based on the statistical and semantic representation of documents. Our work comprises the extraction and filtering of opinions from the text and the assignment of the mark to subjective sentences. The extrac- tion and information filtering consists of the identification of quite precise information in a text in natural language and its representation in a structured form (13).
... Pour classer les opinions nous avons utilisé une notation suivant une échelle variant de 1 à 5. Nous avons utilisé trois différentes approches pour effectuer la notation. Nous présentons les classificateurs linguistiques[Dziczkowski & Wegrzyn-Wolska (2007a)] et le classificateur basé sur le comportement des groupes que nous avons proposés. Ces deux approches sont ensuite comparées avec le classificateur "naïf Bayes" et SVM [Figure 5.7]. ...
This thesis describes the study and development of a system designed for the evaluation of sentiments within cinema reviews. Such a system offers :
- an automatic search of reviews on the Internet,
- the valuation and the attribution of marks to the opinions given by cinema critics,
- the publication of the results.
In order to improve the application results of predictive algorithms the objective of this system is to supply a support system for the prediction engines analysing users profiles. Firstly the system seeks and fetches likely reviews by cinema reviewers on the internet, particularly those who are prolific. Then the system will evaluate and attribute a mark to the opinion expressed in the cinema reviews and automatically associate a numerical mark to each review ; this is the objective of the system. The final stage is to regroup the reviews (as well as the marks) with the user who wrote them so as to create complete profiles and to propose these profiles the prediction engines.
For the development of this system research for this thesis was based principally on the marking of sentiments, this work is in the realm of Opinion Mining and Sentiment Analysis. Our system uses three different methods for the classification of opinions. We present here two new methods ; one founded on pure linguistic knowledge and the other on a combination of statistic and linguistic analysis. Subsequently the results are compared using the statistical method based on Bayes' classifier frequently used in this domain.
The ensuing results are then combined in order to make the final evaluation as precise as possible. For this task we used a fourth classifier based on the neuron network.
Between one and five points are attributed to reviews. This mark requires a deeper linguistic analysis than the binary notation- positive/negative which may be objective or subjective and which is habitually used.
This thesis gives a general account of all the system modules which we have created and a detailed analysis of the one dedicated to opinion marking. We wish to show the dvantages of deep linguistic analysis which is less commonly used than statistical analysis in the domain of sentiment analysis.
... At the end we obtained three marks for one review which can be different. We use another classifier which will assign the final mark to the reviews based only on three marks get before from classifiers (Dziczkowski and Wegrzyn-Wolska, 2007a), (Dziczkowski and Wegrzyn-Wolska, 2007b). ...
... At the end we obtained three marks for one review which can be different. We use another classifier which will assign the final mark to the reviews based only on three marks get before from classifiers (Dziczkowski and Wegrzyn-Wolska, 2007a), (Dziczkowski and Wegrzyn-Wolska, 2007b). The process of assignment of the mark into the critic is shown on figure 1. ...
This paper describes the part of recommender system designed for movies' critics recognition. Such a system allows the automatic collection, evaluation and rating of critics and opinions of the movies. First the system searches and retrieves texts supposed to be movies' reviews from the Internet. Subsequently the system carries out an evaluation and rating of movies' critics. Finally the system automatically associates a numerical mark to each critic. The goal of system is to give the score of critics associated to the users' who wrote them. All of this data are the input to the cognitive engine. Data from our base allow making correspondences which are required for cognitive algorithms to improve advanced recommending functionalities for e-business and e-purchases websites. Our sesystem uses three different methods for classifying opinions from reviews critics. In this paper we describe the part of system which is based on automatically identifying opinions using natural language processing knowledge.
... Pour classer les opinions nous avons utilisé une notation suivant une échelle variant de 1 à 5. Nous avons utilisé trois différentes approches pour effectuer la notation. Nous présentons les classificateurs linguistiques[Dziczkowski & Wegrzyn-Wolska (2007a)] et le classificateur basé sur le comportement des groupes que nous avons proposés. Ces deux approches sont ensuite comparées avec le classificateur "naïf Bayes" et SVM [Figure 5.7]. ...
... These characteristics are for example: a typical word, typical expression, a size of a sentence, the frequency of characteristic word repetition, the number of punctuation marks (!, ;), ?) and so on. For group categorizing, we used a linguistic analyser Unitex, to lemmatize the words, to assign semantic classes to the words, to add synonyms [4] and to detect negation. For this task we used a linguistic processing, which requires lexicons and specialized grammar. ...
... The semantic analysis aims at producing a structure representing, as accurately as possible, a unit of the sentence, with its meanings and its complexity [1][11][6]. Semantico-conceptual structures can be more or less broad, rich and complex and more or less ambiguous [4]. To determine the behaviour of a group we parse the large corpus of reviews , which were assigned with the same mark to find the characteristic. ...
This paper describes the part of a recommendation system designed for the recognition of film reviews (RRSS). Such a system
allows the automatic collection, evaluation and rating of reviews and opinions of the movies. First the system searches and
retrieves texts supposed to be movie reviews from the Internet. Subsequently the system carries out an evaluation and rating
of the movie reviews. Finally, the system automatically associates a digital assessment with each review. The goal of the
system is to give the score of reviews associated with the user who wrote them. All of this data is the input to the cognitive
engine. Data from our base allows the making of correspondences, which are required for cognitive algorithms to improve, advanced
recommending functionalities for e-business and e-purchase websites. In this paper we will describe the different methods
on automatically identifying opinions using natural language knowledge and techniques of classification.
... • Integration Frameworks -integration of TM (ex. The "MedScan" [7] system combines lexicons with syntactic and semantic templates in a general-purpose TM system to extract relationships between biomedical entities). Our work focused on two branches of biomedicine TM, Named Entity Recognition (NER) and Synonym and Abbreviation Extraction. ...
The quantity of biomedical publications is growing at an exponential rate. With such explosive growth of the content, it is more and more difficult to locate, retrieve and manage the resulting information. This is why text mining has become a necessity. The main goal of biomedical research is to put knowledge to practical use in the form of diagnoses, prevention, and treatment. It is important to pool the resources between the different individuals researching results. The objective of this paper is to discuss the variety of issues and challenges surrounding the perspectives regarding the use of Information Retrieval and Text Mining methods in biomedicine. The article will first look at the directions in biomedical TM and then describe the work done for the BIAM project, the French on-line Medical Data Base.
