=Paper=
{{Paper
|id=Vol-1178/CLEF2012wn-PAN-CastilloEt2012
|storemode=property
|title=Graph-based and Lexical-Syntactic Approaches for the Authorship Attribution Task
|pdfUrl=https://ceur-ws.org/Vol-1178/CLEF2012wn-PAN-CastilloEt2012.pdf
|volume=Vol-1178
|dblpUrl=https://dblp.org/rec/conf/clef/CastilloAPOGC12
}}
==Graph-based and Lexical-Syntactic Approaches for the Authorship Attribution Task==
https://ceur-ws.org/Vol-1178/CLEF2012wn-PAN-CastilloEt2012.pdf
Graph-based and Lexical-Syntactic Approaches
for the Authorship Attribution Task
Notebook for PAN at CLEF 2012
Esteban Castillo1 , Darnes Vilariño1 , David Pinto1
Iván Olmos1 , Jesús A. González2 , and Maya Carrillo1
1
Benemérita Universidad Autónoma de Puebla
Faculty of Computer Science, Mexico
2
Institute of Astrophysics, Optics, and Electronics
Computer Science Department, Mexico
ecjbuap@gmail.com,{darnes, dpinto, iolmos, cmaya}@cs.buap.mx
jagonzalez@inaoep.mx
Abstract. In this paper we present two different approaches for tackling the au-
thorship attribution task. The first approach uses a set of phrase-level lexical-
syntactic features, whereas the second approach considers a graph-based text rep-
resentation together with a data mining technique for discovering authorship pat-
terns which may be further used for attributing the authorship of an anonymous
document. In both cases we employed a support vector machine classifier in order
to determine the target class. The features extracted by means of the graph-based
approach allowed it to obtain a better performance than the other approach.
Keywords: Authorship attribution, graph-based representation, phrase-level lexical-
syntactic features, support vector machines
1 Introduction
Discovering the correct features in a raw text which allows unambiguously to attribute
the authorship of a given anonymous document is a very hard task. In recent years, there
have been a number of research papers in this direction. The traditional authorship attri-
bution task consists of determining the correct authorship of an anonymous document,
using a supervised collection of documents, i.e., a reference set of documents manually
tagged with their corresponding authorship attribution. In other words, this task can be
seen as a classification problem in which the target tag or class is the author name/ID.
Determining the authorship of an anonymous document is a task that has been tack-
led for several years by the computational linguistic community. An effort that has been
empowered by the continuous growing of information in Internet. In this sense, the
importance of finding the correct features for characterizing the signature or particular
writing style of a given author is fundamental for solving the problem of authorship
attribution.
The results reported in this paper were obtained in the framework of the 6th Inter-
national Workshop on Uncovering Plagiarism, Authorship, and Social Software Misuse
(PAN’12). In particular, in the task named “Traditional Authorship Attribution” which
has the following two sub-tasks:
– Traditional (closed class / open class, with a variable number of candidate authors).
This subtask consists of assigning the real author name to a given anonymous doc-
ument, using a set of candidate authors as reference.
– Clustering. A target number of paragraphs are required to be clustered into groups
(between one or four) in order to obtain clusters of paragraphs that correspond to
the same author.
For this purpose, we attempted two different techniques for representing the features
that will be taken into account in the process of authorship attribution. The proposed
approaches are discussed in the following sections.
The rest of this paper is structured as follows. In Section 2 it is presented the de-
scription of the features used in the task to be tackled. Section 3 shows the classification
methods (supervised and unsupervised) employed in the experiments. The experimen-
tal setting and a discussion of the obtained results are given in Section 4. Finally, the
conclusions of this research work is presented in Section 5.
2 Description of the features used in the task
In this work we explore two very different text representation schemas. The first ap-
proach considers lexical-syntactic features, whereas the second uses a data mining
based process for extracting the most relevant terms of the target documents. Both
schemas are described as follows.
2.1 Lexical-syntactic feature approach
In this approach are considered the following lexical-syntactic features for representing
the particular writing style of a given author:
– Phrase level features
• Word prefixes. A group of letters added before a word or base to alter its mean-
ing and form a new word.
• Word sufixes. A group of letters added after a word or base to alter its meaning
and form a new word.
• Stopwords. A group of words that bear no content or relevant semantics which
are filtered out from the texts.
• Trigrams of PoS. Sequences of three PoS tags1 appearing in the document.
– Character level features
• Vowel combination. Word consonants are removed and, thereafter, the remain-
ing vowels are combined. Each vowel combination is considered to be a fea-
ture. Adjacent repetition of vowels are considered as only one vowel.
• Vowel permutation. Word consonants are removed and, thereafter, the vowel
permutation is considered to be a feature.
The text representation by means of the above mentioned features is described in
Section 2.3.
1
POS-tagger of NLTK
2.2 Graph-based approach
In this approach, a graph based representation is considered[5]. Given a graph G =
(V, E, L, f ) with V being the non-empty set of vertices, E ⊆ V × V the edges, L the
tag set, and f : E → L, a function that assigns a tag to a pair of associated vertices.
Each text paragraph is tagged with its corresponding PoS tags, in this case, using the
TreeTagger tool2 . Each word is stemmed using the Porter stemmer3 . In this type of
text representation, each vertex is considered to be a stemmed word and each edge is
considered to be its corresponding PoS tag. The word sequence of the paragraphs to be
represented is kept. The tag set of PoS used in the experiments is shown in Table 1.
Table 1. Description of PoS tags used
PoS tag Description
JJ Adjective
VBN Verb - Past participle
WDT Determiner
NN Noun
CD Number
RB Adverb
NNS Noun - Singular
CC Conjuntion
RBR Adverb - Comparative
MD Modal
JJR Adjective - Comparative
VBG Verb - Present participle
VBD Verb - Past
VBP Verb - Present, not the 3rd person singular
VBZ Verb - Present, 3rd person singular
FW Unknown word
PRP Possessive pronoun
VB Verb in base form
NNP Noun - Plural
RBS Adverb - Superlative
IN Preposition and conjunction
JJS Adjective superlative
PDT Predeterminer
In order to demonstrate the way we construct the graph for each phrase, consider
the following text phrase: “second qualifier long road leading 1998 world cup”. The
associated graph representation is shown in Figure 1.
The Subdue tool Once each paragraph is represented by means of a graph, we apply a
data mining algorithm in order to find subgraphs. Subdue is a data mining tool widely
used in structured domains. This tool has been used for discovering structured patterns
in texts represented by means of graphs [2]. Subdue uses an evaluation model named
“Minimum encoding”, a technique derived from the minimum description length prin-
ciple [3], in which the best graph sub-structures are chosen. The best subgraphs are
those that minimize the number of bits that represent the graph. In this case, the number
of bits is calculated considering the size of the graph adjancency matrix. Thus, the best
2
http://www.ims.uni-stuttgart.de/projekte/corplex/TreeTagger/
3
http://tartarus.org/ martin/PorterStemmer/
Fig. 1. Graph based text representation with words and their corresponding PoS tags
substructure is the one that minimizes I(S) + I(G|S), where I(S) is the number of bits
required to describe the substructure S, and I(G|S) is the number of bits required to
describe graph G after it has been compacted by the substructure S.
2.3 Text representation schema
Let (x1 , x2 , x3 , · · · , xn ) be the set of features selected for representing the documents.
Each document D is represented considering the feature frequency, i.e., we used the bag
of words representation for each document[1]. It is worth noting that both approaches
(lexical-syntactic and graph-based) use the same text representation schema.
The training stage uses the following feature vector:
D = (x1 , x2 , x3 , . . . , xn , C) (1)
| {z }
Document f eatures
where C is the class manually associated to the document, in this case, the author Name
or ID.
For the testing stage, we use the feature vector as follows:
D = (x1 , x2 , x3 , . . . , xn ) (2)
| {z }
Document f eatures
In this case, there is not a classification attribute (class name) due to the anonymous
source of the document.
3 Description of the classifiers used in the task
We have used a Support Vector Machine (SVM) classifier for the problems A, B, C, D,
I and J (see Section 4.1). SVM is a learning method based on the use of a hypothesis
space of lineal functions in a higher dimensional space induced by a kernel, in which
the hypotheses are trained by one algorithm that uses elements of the generalization
theory and taken from the optimization theory.
The linear learning machines are barely used in major real world applications due to
their computational limitations. Kernel based representations are an alternative for this
problem proyecting the information to a feature space of higher dimensionality which
increases the computational capacity of the linear learning machines. The input space
X is mapped to a new feature space as follows:
x = {x1 , x2 , . . . , xn } → φ(x) = {φ(x)1 , φ(x)2 , . . . , φ(x)n } (3)
By employing the kernel function, it is not necessary to explicitly calculate the
mapping φ : X → F in order to learn in the feature space.
In this research work, we employed as kernel the polynomial mapping, which is a
very popular method for modeling non-linear functions:
K(x, x) = (hx, xi + c)d (4)
where c ∈ R.
For problems E and F, we have employed the K-means clustering method, repre-
senting the documents with the six lexical-syntactic features previously presented. K-
means is a cluster analysis method that aims to partition n observations into K clusters,
considering that each observation belongs to the cluster with the closest median.
In the experiments carried out in this paper, we used the Weka data mining plat-
form[4] for executing the implementations of SVM and the K-means classifier.
4 Experimental results
The results obtained with both approaches are discussed in this section. First, we de-
scribe the dataset used in the experiments and, thereafter, the obtained results.
4.1 Data sets
The description of the eight text collections used in the experiments (six for the tra-
ditional sub-task and two for the clustering sub-task) is shown in Table 2. As can be
seen, the data set is made up of different authors. Actually, the first and second text
collections (A and B) contain three different authors, the third and fourth collections (C
and D) contain eight different authors, the fifth and sixth collections (I and J) contain
14 different authors, and finally, the seventh and eighth collections (E and F) contain
mixed and intrusive paragraphs from 1 to 4 different authors.
4.2 Results obtained in the traditional sub-task
In Table 3 are shown the results obtained for the problems A, B, C, D, I and J of the
traditional sub-task.
In order to tackle the open-class problems (B, D, and J), the training data set was
enriched with documents written by unknown authors. The number of documents added
Table 2. Data set used in the experiments
Task Problem type Authors Documents training Documents test
Traditional A closed-class 3 6 6
Traditional B open-class 3 6 10
Traditional C closed-class 8 16 8
Traditional D open-class 8 16 17
Traditional I closed-class 14 28 14
Traditional J open-class 14 28 16
clustering E mixed paragraph documents 1-4 2(6 paragraphs) 3 (90 paragraphs)
clustering F intrusive paragraph documents 1-4 2(17 paragraphs) 4(80 paragraphs)
Table 3. Results obtained in the traditional sub-task
Task A correct/A% B correct/B% C correct/C% D correct/D% I correct/I% J correct/J%
Graph-based approach 5/83.333 6/60 5/62.5 4/23.529 8/57.142 13/81.25
Lexical-syntactic approach 4/66.666 3/30 2/25 6/35.294 10/71.428 7/43.75
was exactly the same that the number of documents of each original collection. As may
be seen in the obtained results, the best results were obtained with the graph-based
representation, in which the best features were discovered with the Subdue tool. The
closed-class problems obtained a better performance than the open-class ones which
encourages us to investigate better methods for tackling this particular issue. It is worth
noting that we retrieved almost all the authors for the open-class problem J. We consider
that the number of training data is an important factor on this behavior, in other words,
we consider increasing the amount of information provided by the authors.
4.3 Results obtained in the clustering sub-task
Table 4 shows the results obtained in the problems E and F of the clustering sub-task.
Table 4. Results obtained in the clustering sub-task
Task E correct/E% F correct/F%
Graph-based approach 68/75.555 43/53.75
Lexical-Syntactic approach 61/67.777 51/63.75
Different runs varying the number of clusters (K) were sent to the competition,
however, we are not aware of the final run (the number K) reported in this paper, which
was evaluated by the competition organizers. The different runs were motivated by an
empirical value selected by means of the cosine similarity among different paragraphs
of the text collection. This metric was a clue for determining the final number of clus-
ters.
5 Conclusions
In this paper we presented an exploration of two different approaches. On the first hand,
we employed a graph for representing text paragraphs by means of words and their cor-
responding part of speech tags. We aimed to consider the morphosyntactical structure
of the text (at once) for further discovering of the best features for the final repre-
sentation of the training and test data. On the other hand, we evaluated a set of six
lexical-syntactic features with the purpose of determining those that allow finding an
appropriate signature for a given author. A higher number of features (phrase and word
level) were independently evaluated, and those that provided the best discrimination
scores were selected for the final evaluation.
In general, we observed that the graph-based representation obtained a better per-
formance than the other one. However, more investigation on the graph representation
is still required, so that graph patterns discovered by the Subdue tool are better than the
ones obtained until now. As future work, we want to experiment with different graph-
based text representations that allow us to obtain much more complex patterns.
References
1. Manning, C.D., Raghavan, P., Schtze, H.: Introduction to Information Retrieval. Cambridge
University Press, New York, NY, USA (2008)
2. Olmos, I., Gonzalez, J.A., Osorio, M.: Subgraph isomorphism detection using a code based
representation. In: FLAIRS Conference. pp. 474–479 (2005)
3. Rissanen, J.: Stochastic Complexity in Statistical Inquiry Theory. World Scientific
Publishing Co., Inc., River Edge, NJ, USA (1989)
4. Witten, I.H., Frank, E.: Data Mining: Practical Machine Learning Tools and Techniques.
Morgan Kaufmann Series in Data Management Sys, Morgan Kaufmann, second edn. (June
2005)
5. Yan, X., Han, J.: gspan: Graph-based substructure pattern mining. In: ICDM. pp. 721–724
(2002)