程序代写代做代考 data mining database prolog algorithm untitled

untitled

A Frequent Keyword-set based Algorithm for Topic
Modeling and Clustering of Research Papers

Kumar Shubhankar

Centre for Data Engineering
IIIT Hyderabad

Hyderabad, India

shubankar@students.iiit.ac.in

Aditya Pratap Singh
Centre for Data Engineering

IIIT Hyderabad
Hyderabad, India

aditya_pratap@students.iiit.ac.in

Vikram Pudi
Centre for Data Engineering

IIIT Hyderabad
Hyderabad, India

vikram@ iiit.ac.in

Abstract – In this paper we introduce a novel and efficient
approach to detect topics in a large corpus of research
papers. With rapidly growing size of academic literature, the
problem of topic detection has become a very challenging
task. We present a unique approach that uses closed
frequent keyword-set to form topics. Our approach also
provides a natural method to cluster the research papers
into hierarchical, overlapping clusters using topic as
similarity measure. To rank the research papers in the topic
cluster, we devise a modified PageRank algorithm that
assigns an authoritative score to each research paper by
considering the sub-graph in which the research paper
appears. We test our algorithms on the DBLP dataset and
experimentally show that our algorithms are fast, effective
and scalable.

Keywords. Closed Frequent Keyword-set, Topic Detection,
Graph Mining, Citation Network, Authoritative Score

I. INTRODUCTION
Thousands of research papers are published every
year spanning various fields of research. These research
papers cover broad topics like Data Mining, Artificial
Intelligence, Information Extraction, etc., which can be
further divided into several sub-topics, sub-sub-topics and
so on. Classification of research papers into the broad
topics is a trivial problem which is generally dealt
manually. But more often than not, a researcher is more
interested in literature related to specific topics within
these broader topics. For example, a researcher might be
more interested in knowing about topics like Data
Visualization, Frequent Pattern Mining, etc., which are
different sub-topics within the broader field of Data
Mining. Topic discovery has recently attracted
considerable research interest [13, 15, 10, 21, 22, 24, 19,
4, 18, 8, 17]. Also, the researchers are often interested in
knowing the important research papers from their topics
of interest.
In this paper, we propose a novel and efficient
method for topic modeling and clustering of research
papers. The keywords associated with a paper can be used
to deduce the topics the paper deals with. Often these
keywords are fed as meta-data by the authors in the paper
itself. But this is not the case for all the research papers.
We need to have a deterministic way of determining the
topics of concern for all the research papers. Based on the
intuition that a document is well summarized by its title
and the title gives a fairly good high-level description of

its content, we use the keywords present in the title of a
paper to detect the topics. It is to be noted that we do not
use abstract of the paper for extracting phrases as there
are lots of irrelevant phrases in the abstract present as
noise.
In the proposed approach, we form closed frequent
keyword-sets by top-down dissociation of keywords from
the phrases present in the titles of papers on a user-
defined minimum support. All the research papers sharing
a topic form a natural cluster. We also propose a time
independent, modified iterative PageRank algorithm to
rank the research papers in a topic cluster which assign an
authoritative score to each paper based on the citation
network. For a topic T we consider all the research papers
containing that topic and the citation edges of these
papers. We can determine the important research paper
belonging to a given topic T by looking at the scores of all
the research papers belonging to that topic. It is to be
noted that a research paper could belong to a number of
clusters forming hierarchical, overlapping clusters.
These algorithms have many applications like
recommendation systems, finding landmark papers, trend
analysis etc. We test our algorithms on the DBLP dataset.
Our experiments produce a ranked set of research papers
corresponding to each topic that on examination by field
experts and based on our study match the prominent
papers from the topics in the dataset.

II. RELATED WORK
Topic extraction from documents has been studied
by a lot of researchers. Most work on topic modeling is
statistics-based like the work by Christain Wartena and
Rogier Brusse [23], which uses most frequent nouns, verbs
and proper names as keywords and clusters them based on
different similarity measures using the induced k-bisecting
clustering algorithm. There has also been some NLP-based
work on topic detection like the work by Elena Lloret [16].
Our work is based on dissociation of phrases into frequent
keyword-sets, which as shown in section 7 is very fast and
highly scalable.
Clustering documents based on frequent itemsets
[2] has been studied in the algorithms FTC and HFTC [3]
and the Apriori-based algorithm [14]. Both of these works
consider the documents as bags of words and then find
frequent itemsets. Thus, the semantic information present
in the document is lost. We extract phrases from the titles
of the research papers and derive frequent substrings as

frequent keyword-sets, maintaining
semantics. Another work that talks abou
of semantics while forming topics is by
[6], where the authors consider a wind
they find the itemsets which are candidat
within this window, the relative positio
considered insignificant. We, on the
considered only sequential patterns with
quite intuitive that if we do not con
positions of the keywords, there might b
semantics are lost.
We have used closed [20] freque
similarity measure rather than maximal
set as used by Ling Zhuang et al. [25]
document clustering. We cannot use
keyword-sets as topics because the
information is lost as it considers only th
keyword-set.
Topic summarization and analy
documents has been studied by Xueyu
Wang [9]. They have used LDA mode
and Kullback-Leibler divergence as simi
clustering the documents. The drawback
that it needs a pre-specified number of
manual topic labeling. In our study, no p
topics is required. The work in [11] us
between the distribution of terms repres
the distribution of links in the citation
documents containing these terms. We o
use frequent keyword-sets to form the top
citation links to detect important topics
derived.
Rest of the paper is organized as
describes our topic modeling algor
discusses our approach to cluster the re
topics. Section 5 describes our algorith
research papers using citation network.
the experiments and results. Section 7
applications of our algorithms and sectio
work.

III. OUR APPROAC
The method proposed by us i
formation of keyword-sets from titles
papers and finding closed frequent keyw
the topics.

Definition 1. Phrase: A phrase P is def
words between two stop-words in the ti
paper.

Definition 2. Keyword-set: We define a
an n-gram substring of a phrase, wher
number.

Definition 3. Frequent Keyword-set: A
said to be frequent if its count in the c
than or equal to a user-defined minimum

the underlying
ut the preservation
Zhou Chong et al.
dow within which
tes for topics. But,

on of the words is
other hand, have
hin a phrase. It is

nsider the relative
be cases where the

ent keyword-set as
frequent keyword-
] in their work on
maximal frequent

en most of the
he longest possible

ysis on academic
Geng and Jinlong
l to extract topics
ilarity measure for

k of LDA model is
f latent topics and
prior knowledge of
ses the correlation
senting a topic and

graph among the
on the other hand,
pics and utilize the
among the topics

follows: Section 3
rithm. Section 4
search papers into
hm for ranking of
Section 6 presents
talks about some

on 8 concludes our

H
is based on the

of the research
word-sets to form

fined as a run of
itle of a research

keyword-set K as
re n is a positive

keyword-set K is
corpus is greater

m support [1].

Definition 4. Closed Frequent Key
closed frequent keyword-set as a
none of whose supersets has the sa
papers as it. Each closed frequent k
a unique topic T.

A. Phrase Extraction
Given the title of a research p
its phrases Pij. As defined above,
words between two stop-words from
of 671 Standard English stop-word
is thus mapped to the corresponding
title.
Our original problem domain

R1 – [P11, P12, P13
R2 – [P21, P22, P23

Here, Ri represents the i
th r

represents its jth phrase. Our ne
keyword-sets from these phrases, di
into keyword-sets. It should be not
problem domain, we will have t
phrase multiple times in a single s
occur in several research papers. S
problem domain, mapping each p
research papers Rij it belongs to
dataset. In this domain, each phra
only once. The problem domain
follows:

P1 – [R11, R12, R13
P2 – [R21, R22, R23

B. Keyword-set Extraction
As defined above, a keywor
substring of a phrase P. In our method,
the substrings of the phrases as keyw
relative ordering is maintained, pre
semantic of the phrases. Each keywo
semantic unit that can function as a
knowledge discovery and hence is a pot
As an example of keyword-set
phrase ABCDE, the potential frequent k
all the ordered substrings. Thus, ABC
keyword-sets, as shown in figure 1.

Figure 1. Set of all keyword-sets obtaine

Note that ABCDE is a phr
keywords A, B, C, D and E. Find

yword-set: We define a
frequent keyword-set

ame cluster of research
keyword-set represents

paper Ri, we extract all
a phrase is a run of
m a comprehensive list

ds. Each research paper
g phrases present in its

n is:

3, …]
3, …]

research paper and Pij
ext step is to derive
issociating each phrase
ted that in the original
to dissociate a given

scan as a phrase might
o, we reverse map the
hrase Pi to the set of
, in one scan of the

ase will be dissociated
is thus modified as

3, …]
3, …]

rd-set K is basically a
we have considered only

word-sets and hence the
eserving the underlying
ord-set thus formed is a

basic building block of
tential topic.

extraction, consider the
keyword-sets are the set of
CDE gives the following

ed from phrase ABCDE

rase consisting of the
ding all the substrings

requires a simple implementation of queue in top-down
fashion, taking O(1) time at each level and O(n) time
overall. We consider only the substrings of a phrase rather
than the power set of the keywords in the phrase. Thus,
finding the keyword-sets requires O(n) time instead of
O(2n).

C. Frequent Keyword-set Formation
Frequent keyword-sets are formed on a user-
defined minimum support. Frequent keyword-sets are
those keyword-sets whose support is no less than the
minimum support. The supports of the keyword-sets are
calculated during the generation of the keyword-sets from
the phrases in the second scan.
It is to be noted that the length of the list of
research papers corresponding to a phrase is its support,
assuming that a phrase occurs not more than once in the
title of the research paper. In the first scan, we cannot
eliminate the phrases whose support is less than the
minimum support as two or more phrases can share the
same keyword-set whose combined support might be
greater than the minimum support. The elimination of
non-frequent keyword-sets would be done only after all
the keyword-sets, along with their supports, have been
generated in the second scan of the dataset.
The algorithm to increment the support and add
research papers to a given keyword-set is shown below:

Procedure 1: Frequent Keyword-set Generation

Require: phraseKeys PK, minimum support min_sup

1: for each phrase P in PK
2: keywordSetList KSL= findAllSubstringOf(P)

3: for each keywordSet K in KSL
4: keywordSetCount[K] += 1;
5: add paper R to keywordSetPaperList[K]

6: for each keywordSet K in keywordSetCount
7: if keywordSetCount[K] < min_sup 8: delete(keywordSetCount[K]) 9: delete(keywordPaperList[K]) In the procedure 1, all the frequent keyword-sets are derived along with their supports. From step 1 to step 5, all the keyword-sets of each phrase are derived and their supports in keywordSetCount and the corresponding paper list in keywordSetPaperList are updated. From step 7 to step 10, we delete those keyword-sets whose supports are less than the minimum support. The keyword-sets thus derived are the frequent keyword-sets. We have derived the frequent keyword-sets in a top-down fashion. As said before, we only need the substrings of the phrases and not the subsets, which would take O(2n) time. Traditional association rule mining algorithms like Apriori that require one scan of the dataset to calculate the supports of the itemsets at each level take too much time and space. In our algorithm, we require only 2 scans of the dataset to calculate the supports of all the candidate keyword-sets. Since our algorithm runs in linear time compared to exponential Apriori-like algorithms and takes only 2 scans of the dataset to calculate the supports, our algorithms are fast and highly scalable, as shown experimentally in section 6. Also, in Apriori-like algorithms which build higher length itemsets from smaller ones, the relative ordering between the itemsets is lost. In our method, we have considered only the substrings of the phrases as keyword-sets and hence the relative ordering is maintained preserving the underlying semantic of the phrases. D. Closed Frequent Keyword-sets as Topics At this point, we have the frequent keyword-sets. The papers that share the same keyword-set lie in the same cluster. In our algorithm, we may derive non-closed frequent keyword-sets as well. For example, ABCD and ABC may have the same list of papers in their clusters. In this case, we can remove ABC as it does not have any information that ABCD does not have. Our topic should consist of the maximal number of common keywords present in all the papers in the cluster. Thus we need to have closed frequent keyword-sets as topics. Notice that we calculated the support of the frequent keyword-sets in top-down fashion in a single scan in procedure 1. So we cannot eliminate the non-closed keyword-sets in procedure 1 itself. To eliminate the non-closed frequent keyword-sets, we iterate level-wise in the list of frequent keyword-sets. We store the frequent keyword-sets in a level-wise manner, with the number of keywords in the keyword-set representing its level. For every keyword-set of length i, we iterate over the list of keyword-sets of length (i+1) and if i-length keyword-set is a substring of (i+1)-length keyword-set and the support is same for both, we delete the i-length keyword-set, as it is non-closed. IV. CLUSTERING RESEARCH PAPERS BASED ON TOPICS Till now, we have closed frequent keyword-sets, each representing a cluster of research papers. These topic clusters are complete in the sense that we have the maximal length keyword-set shared by all the research papers represented by that topic. In the mapping Keyword Set Paper List, we have the list of papers corresponding to a topic. These sets of papers in the list form different clusters. These clusters of research papers are overlapping in nature as a paper may span more than one topic. These clusters are also hierarchical in nature. The cluster representing a broader topic is essentially a combination of several clusters representing its sub-topics, which in turn are a combination of the sub-sub-topic clusters and so on. For example, databas system is a broad topic and imag databas systems, distributed databas systems, etc. are sub-topics lying within the broader topic databas system. Each level of the hierarchy represents a different level of data description, which facilitates the knowledge discovery at various levels of abstraction. Thus, we have used closed frequent keyword-sets to form topics and used these topics as the similarity measure to cluster the research papers. V. RANKING OF RESEARCH PAPERS For each topic we have a cluster of research papers in which the topic lies. To find out which research papers are of good quality, we have developed a time independent, modified iterative PageRank algorithm. We ranked the research papers based on citation network. Each research paper is cited by a number of research papers and there exists a well-defined graph structure among the network of research papers. The basic algorithm for ranking the research papers based on citation network uses the two types of edges in a graph: Outlinks and Inlinks. Definition 5. Outlinks: From a given node N, link all the nodes Ni that the node N cites. Definition 6. Inlinks: To a given node N, link all the nodes Nj that cite the node N. These outlinks and inlinks will be used while calculating the authoritative score [12] for each node. The procedure 2 uses modified iterative PageRank algorithm to calculate the authoritative score for each node. Procedure 2: Time-independent Modified Iterative PageRank Algorithm Require: Citation Network CN, Paper Year PY, Outlinks Count OC, Paper Inlinks PI, Damping Factor θ 1: for each paper R in PY 2: year Y = PY[R] 3: Year Citation Count YCC[Y] += OC[R] 4: Year Paper Count YPC[Y] += 1 5: for each year Y in YCC 6: Average Year Citations Count AYCC[Y] = YCC[Y]/YPC[Y] 7: Initialize Paper Rank PR to 1.0 for each paper R 8: while true 9: flag = true 10: for each paper R in PR 11: Current Score CS = PR[R] 12: if R in PI 13: Inlinks List IL = PI[R] 14: New Score NS = 0.0 15: for each inlink I in IL 16: if I in PR 17: NS += PR[I]/OC[I] 18: year Y = PY[R] 19: NS = (1-θ) + θ*NS/AYCC[Y] 20: if CS is not equal to NS 21: flag = false 22: Updated Paper Rank UPR[R] = NS 23: if flag is equal to true 24: break 25: copy UPR to PR 26: clear UPR 27: Maximum Score MS = Maximum Score in PR 28: for each paper R in PR 29: PR[R] /= MS In the procedure 2, we have the citation network CN as input from which we create an Outlink Count OC for each research paper which keeps track of the number of outlinks corresponding to each research paper. We also create Paper Inlinks PI which maps each paper to the list of papers which are its inlinks. In the PageRank algorithm, the damping factor θ is also used which is required to prevent the scores of research papers that do not have any inlinks from falling to zero. For the experiments we set the damping factor to 0.85 [5] which gave satisfactory result. In steps 1-6, we calculate the value of metric Average Year Citations Count AYCC which is the metric we introduce to counter the time dependence of PageRank algorithm. This metric is the average number of citations per paper in a particular year which is a time dependent metric and directly reflects the varying distribution of citations over the years. We observe that this metric captures the time bias against the newer papers well and has high values for older papers and low values for newer ones. Considering the year of publication of all the research papers, we pre- compute the total number of citations for each year and the number of research papers published in each year. Using them, the average number of citations per paper for each year is determined. In line 7, each paper’s score is initialized to unity and then the algorithm iteratively modifies the score depending on the score of other papers that point towards it. It stops when all the research paper’s scores converge, i.e. become constant. Step 8 to step 26 is the iterative calculation of the authoritative score for each research paper R. The iteration stops when there is no change in the score of any R. This is signified by no change in the value of flag, set to true in step 9; flag is set to false in step 16 if there is a change in the score of any paper R during the update step. From step 10 to step 22, for each research paper, a new authoritative score is calculated based on the scores of the inlinks in the previous iteration. The PageRank algorithm is based on the fact that the quality of a node is equivalent to the summation of the qualities of the nodes that point to it. Here, quality refers to the score of the research paper. This fact is used in step 17 by dividing the inlink score by OC[I] which is the number of outlinks of the inlink. This takes care of the fact that if a research paper cites more than one paper, it depicts that it has drawn inspiration from various sources and hence its effect on the score of the paper it cites should diminish by a factor equal to the number of paper it cites. Step 19 modifies the score calculated above to incorporate the damping factor θ. It also incorporates the time-independence factor by using the average citations per paper in a year metric AYCC[Y]. In step 23-24, if the scores of none of the papers differ from the previous iteration, the value of flag is unchanged from its true value, and the loop breaks indicating the convergence of the algorithms. In step 27 to 29, we normalize the scores to scale down the scores within the range [0, 1]. Thus, finally we get the authoritative score of each research paper R based on the score of the research papers citing it. VI. EXPERIMENTS AND RESULTS A. Dataset To show the results of our algorithms, we used the DBLP XML Records available at http://dblp.uni- trier.de/xml/ [7]. The DBLP dataset contains information about various research papers from various fields published over the years. This information includes the title of the research paper, its author(s), the year of publication, the conference of publication, a unique key for each research paper and the keys of the research papers the given research paper cites. It is to be noted that the dataset used by us contained research papers with citation information till the year 2010 only. B. Data Preprocessing The DBLP dataset also contains information that is not useful in our algorithms. We need to pre-process the dataset to extract only the information that we will use in our algorithms. In data pre-processing, we extracted all the research paper titles, year of publication and citations. Also, all the keywords present in the titles of the research papers were stemmed using the Porter's Stemming Algorithm. C. Results We implemented our algorithms on the DBLP dataset and discovered various interesting results. It contains 16,32,442 research papers from various fields of research. An objective and quantitative evaluation of the result is difficult due to the lack of standard formal measures for topic detection tasks. But, the list of topics generated by our experiments on examination by field experts and based on our observations match prevailing topics in the dataset. D. Topic Modeling The topic modeling algorithm consisted of forming closed frequent keyword-sets extracted from the phrases present in titles of the research papers in the DBLP dataset. Each of the closed frequent keyword-sets represents a distinct topic. We tested our algorithms on various values of minimum support. Upon implementing the algorithms with minimum support 100, we obtained 12,057 topic clusters consisting of 5,476 1-length topics, 5,766 2- length topics, 748 3-length topics, 62 4-length topics and 5 5-length topics. Based on their support, the top three n- length topics were: TABLE I: TOP 3 n-LENGTH TOPICS BASED ON SUPPORT n-length Top Three Topics 1-length system model network 2-length neural network real time sensor network 3-length wireless sensor network support vector machin ad hoc network 4-length mobil ad hoc network wireless ad hoc network content base imag retriev 5-length radial basi function neural network ieee 802 11 wirelss lan low density pariti check code D.1. Top Research Papers in a Topic Cluster For each topic T, we have the ranked list of research papers as determined by our algorithms. This can be very useful to researchers studying a given topic T. Following is the list of the top three research papers for some of the prominent topics of research. TABLE II: TOP 3 RESEARCH PAPERS FOR SOME PROMINENT TOPICS ALONG WITH THE TOPIC CLUSTER SIZE Topic Cluster Size Top 3 Research Papers algorithm 81506 Introduction to Algorithms. Graph-Based Algorithms for Boolean Function Manipulation. Fast Algorithms for Mining Association Rules in Large Databases. associ rul 1737 Mining Association Rules between Sets of Items in Large Databases. Fast Algorithms for Mining Association Rules in Large Databases. Fast Discovery of Association Rules. onlin social network 104 A familiar face(book): profile elements as signals in an online social network. Information revelation and privacy in online social networks. Measurement and analysis of online social networks. neural network 19178 Neural Networks and the Bias/Variance Dilemma. Image processing with neural networks - a review. Evolving Neural Network through Augmenting Topologies. xml databas 226 TIMBER: A native XML database. Efficient Keyword Search for Smallest LCAs in XML Databases. Querying Structured Text in an XML Database. The top research papers for the t to match the most popular research pape as determined by field experts. Also found to be considerably consistent ir cluster size or the age of the topics. E. Performance To evaluate the performance of o varied minimum support and plotted the algorithm to run. The algorithm bui keyword-sets and eliminates the non-clos The following graph shows this variation Figure 2. Graph showing minimum support versu algorithm to run Above algorithm was run on a 3G 2.10GHz Intel® Core™ 2 Duo CPU. VII. APPLICATIONS Our algorithms have a variety o various fields of research. Following applications areas: • Topic Search System: This system the top ranked papers of a g retrieval of papers on year-wise be given as an option. • Topic Ranking System: Using research papers for the top authoritative scores, we would system that would compare the to a ranked list of topics. • Evolution of Topics: Evolution of tool for new researchers. He can i the emerging fields of research by in the currently popular topics. • Recommendation Systems: Extend to authors and conferences based build a recommendation system. B topics of interests of the a recommend those topics to an other similar authors have worke topics were found ers from that topic our results were rrespective of the our algorithm, we time taken by our ilds the frequent sed frequent ones. n. us time taken by the GB machine with S of applications in are some of the m would retrieve given topic. The granularity could the cluster of pics and their to like build a opics and produce f topics is a useful intuitively deduce y seeing the trends ding the clustering on topics, we can By comparing the authors, we can author on which ed. This similarity between authors could be number of common topics them. Similarly, we can conferences of interest to the • Finding Landmark Papers: contains all the research pa topic, we can determine the for the topic by observing the the topic. A sudden peak suggest the presence of an im topic. These important paper by comparing the authoritativ in the corresponding topic classified as Landmark Pape VIII. CONCLUSIONS AND In this paper, we proposed topics and cluster research papers topics were identified by form keyword-sets as proposed by our alg better than traditional approaches proposed a method to produce the papers within a topic cluster. We a topic modeling as well as the algorithms on varying the minimum As mentioned above, our alg of applications. In future, we would algorithms to build systems for topi and recommendations. We would statistical approaches for topic co other domains like web site clustering, etc. in which our algorith IX. REFEREN [1] R. Agarwal, T. Imielinski, an Association Rules between S Databases,” Proceedings SIGMOD Conference, 1993. [2] R. Agarwal, and R. Srikant, Mining Association Rules,” 20th VLDB Conference, 199 [3] F. Beil, M. Ester, and X. Based Text Clustering,” P International Conference on and Data Mining (KDD), 200 [4] D. M. Blei, and J. D. Laffe Models,” NIPS, 2005. [5] S. Brin, and L. Page, “The Scale Hypertextual Web Proceedings of the 7th Intern World Wide Web, 1998. [6] Z. Chong, L. Yansheng, Z “FICW: Frequent Itemset B with Window Constraint,” Natural Sciences, Vol: 11, N 2006. [7] The DBLP Computer S http://dblp.uni-trier.de/ based on a threshold of research shared by also recommend the authors. Since a topic cluster apers published on the most important papers e graph of evolution of in the graph would mportant paper for the rs could be recognized ve scores of the papers cluster and then be rs. FUTURE WORK d a method to derive into these topics. The ming closed frequent gorithms, which works like Apriori. We also ranked list of research analyzed the results of performance of our m support parameter. gorithms have a variety d like to implement our ic search, topic ranking also like to examine orrelation and explore clustering, document hms can be applied. NCES nd A. Swami, “Mining Sets of Items in Large of the 1993 ACM , “Fast Algorithms for ” Proceedings of the 94. Xu, “Frequent Term- roceeding of the 8th Knowledge Discovery 02. rty, “Correlated Topic Anatomy of a Large- b Search Engine,” national Conference on Z. Lei, and H. Rong, Based Text Clustering ” Wuhan Journal of No: 5, pp: 1345-1351, Science Bibliography. [8] E. Erosheva, S. Fienberg, and J. Lafferty, “Mixed- membership Models of Scientific Publications,” Proceedings of the National Academy of Sciences, 2004. [9] X. Geng, and J. Wang, “Toward theme development analysis with topic clustering,” Proceedings of the 1st International Conference on Advanced Computer Theory and Engineering, 2008. [10] T. I. Griffiths, and M. Steyvers, “Finding Scientific Topics,” Proceedings of the National Academy of Sciences, 2004. [11] Y. Jo, C. Lagoze, and C. L. Giles, ”Detecting Research Topics via the Correlation between the Graphs and Texts,” Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining, 2007. [12] J. Klienberg, “Authoritative sources in a hyperlinked environment,” Proceedings of the 9th Annual ACM-SIAM Symposium on Discrete Algorithms, 1998. [13] J. Kleinberg, “Bursty and Hierarchical Structure in Streams,” Proceedings of SIGKDD, 2002. [14] S. M. Krishna, and S. D. Bhavani, “An Efficient Approach for Text Clustering Based on Frequent Itemsets,” European Journal of Scientific Research, 2010. [15] C.Y. Lin, and E. Hovy, “The Automated Acquisition of Topic Signatures for Text Summarization,” Proceedings of the COLING Conference, 2002. [16] E. Lloret, “Topic Detection and Segmentation in Automatic Text Summarization,” http://www.dlsi.ua.es/~elloret/publications/SumTo pics.pdf, 2009. [17] G. S. Mann, D. Mimmo, and A. McCallum, “Bibliometric Impact Measures Leveraging Topic Analysis,” JCDL, 2006. [18] A. McCallum, A. Corrada-Emmanuel, and X. Wang, “The Author-recipient-topic Model for Topic and Role Discovery in Social Networks: Experiments with Enron and Academic E-mail,” Technical Report, 2004. [19] Q. Mei, and C. Zhai, “Discovery Evolutionary Theme Patterns from Text – An Exploration of Temporal Text Mining,” Proceedings of SIGKDD, 2005. [20] N. Pasquier, Y. Bastide, R. Taoull, and L. Lakhal, “Efficient Mining of Association Rules Using Closed Itemset Lattices,” Information Systems, 1999. [21] M. Steyvers, P. Smyth, M. Rosen-Zvi, and T. I. Griffiths, “Probabilistic Author-topic Models for Information Discovery,” Proceedings of SIGKDD, 2004. [22] X. Wang, and A. McCallum, “Topics over Time: A Non-Markov Continuous-time Model for Topical Trends,” Proceedings of SIGKDD, 2006. [23] C. Wartena, and R. Brussee, “Topic Detection by Clustering Keywords,” Proceedings of the 19th International Conference on Database and Expert Systems Applications, 2008. [24] D. Zhou, E. Manavoglu, J. Li, C. L. Giles, and H. Zha, “Probabilistic Models for Discovering E- communities,” Proceedings of WWW, 2006. [25] L. Zhuang, and H. Dai, “A Maximal Frequent Itemset Approach for Web Document Clustering,” Proceedings of the 4th International Conference on Computer and Information Technology, 2004. << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles true /AutoRotatePages /None /Binding /Left /CalGrayProfile (Gray Gamma 2.2) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Warning /CompatibilityLevel 1.7 /CompressObjects /Off /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.0000 /ColorConversionStrategy /LeaveColorUnchanged /DoThumbnails true /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams true /MaxSubsetPct 100 /Optimize true /OPM 0 /ParseDSCComments false /ParseDSCCommentsForDocInfo false /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo false /PreserveFlatness true /PreserveHalftoneInfo true /PreserveOPIComments false /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Remove /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true /AbadiMT-CondensedLight /ACaslon-Italic /ACaslon-Regular /ACaslon-Semibold /ACaslon-SemiboldItalic /AdobeArabic-Bold /AdobeArabic-BoldItalic /AdobeArabic-Italic /AdobeArabic-Regular /AdobeHebrew-Bold /AdobeHebrew-BoldItalic /AdobeHebrew-Italic /AdobeHebrew-Regular /AdobeHeitiStd-Regular /AdobeMingStd-Light /AdobeMyungjoStd-Medium /AdobePiStd /AdobeSansMM /AdobeSerifMM /AdobeSongStd-Light /AdobeThai-Bold /AdobeThai-BoldItalic /AdobeThai-Italic /AdobeThai-Regular /AGaramond-Bold /AGaramond-BoldItalic /AGaramond-Italic /AGaramond-Regular /AGaramond-Semibold /AGaramond-SemiboldItalic /AgencyFB-Bold /AgencyFB-Reg /AGOldFace-Outline /AharoniBold /Algerian /Americana /Americana-ExtraBold /AndaleMono /AndaleMonoIPA /AngsanaNew /AngsanaNew-Bold /AngsanaNew-BoldItalic /AngsanaNew-Italic /AngsanaUPC /AngsanaUPC-Bold /AngsanaUPC-BoldItalic /AngsanaUPC-Italic /Anna /ArialAlternative /ArialAlternativeSymbol /Arial-Black /Arial-BlackItalic /Arial-BoldItalicMT /Arial-BoldMT /Arial-ItalicMT /ArialMT /ArialMT-Black /ArialNarrow /ArialNarrow-Bold /ArialNarrow-BoldItalic /ArialNarrow-Italic /ArialRoundedMTBold /ArialUnicodeMS /ArrusBT-Bold /ArrusBT-BoldItalic /ArrusBT-Italic /ArrusBT-Roman /AvantGarde-Book /AvantGarde-BookOblique /AvantGarde-Demi /AvantGarde-DemiOblique /AvantGardeITCbyBT-Book /AvantGardeITCbyBT-BookOblique /BakerSignet /BankGothicBT-Medium /Barmeno-Bold /Barmeno-ExtraBold /Barmeno-Medium /Barmeno-Regular /Baskerville /BaskervilleBE-Italic /BaskervilleBE-Medium /BaskervilleBE-MediumItalic /BaskervilleBE-Regular /Baskerville-Bold /Baskerville-BoldItalic /Baskerville-Italic /BaskOldFace /Batang /BatangChe /Bauhaus93 /Bellevue /BellGothicStd-Black /BellGothicStd-Bold /BellGothicStd-Light /BellMT /BellMTBold /BellMTItalic /BerlingAntiqua-Bold /BerlingAntiqua-BoldItalic /BerlingAntiqua-Italic /BerlingAntiqua-Roman /BerlinSansFB-Bold /BerlinSansFBDemi-Bold /BerlinSansFB-Reg /BernardMT-Condensed /BernhardModernBT-Bold /BernhardModernBT-BoldItalic /BernhardModernBT-Italic /BernhardModernBT-Roman /BiffoMT /BinnerD /BinnerGothic /BlackadderITC-Regular /Blackoak /Bodoni /Bodoni-Bold /Bodoni-BoldItalic /Bodoni-Italic /BodoniMT /BodoniMTBlack /BodoniMTBlack-Italic /BodoniMT-Bold /BodoniMT-BoldItalic /BodoniMTCondensed /BodoniMTCondensed-Bold /BodoniMTCondensed-BoldItalic /BodoniMTCondensed-Italic /BodoniMT-Italic /BodoniMTPosterCompressed /Bodoni-Poster /Bodoni-PosterCompressed /BookAntiqua /BookAntiqua-Bold /BookAntiqua-BoldItalic /BookAntiqua-Italic /Bookman-Demi /Bookman-DemiItalic /Bookman-Light /Bookman-LightItalic /BookmanOldStyle /BookmanOldStyle-Bold /BookmanOldStyle-BoldItalic /BookmanOldStyle-Italic /BookshelfSymbolOne-Regular /BookshelfSymbolSeven /BookshelfSymbolThree-Regular /BookshelfSymbolTwo-Regular /Botanical /Boton-Italic /Boton-Medium /Boton-MediumItalic /Boton-Regular /Boulevard /BradleyHandITC /Braggadocio /BritannicBold /Broadway /BrowalliaNew /BrowalliaNew-Bold /BrowalliaNew-BoldItalic /BrowalliaNew-Italic /BrowalliaUPC /BrowalliaUPC-Bold /BrowalliaUPC-BoldItalic /BrowalliaUPC-Italic /BrushScript /BrushScriptMT /CaflischScript-Bold /CaflischScript-Regular /Calibri /Calibri-Bold /Calibri-BoldItalic /Calibri-Italic /CalifornianFB-Bold /CalifornianFB-Italic /CalifornianFB-Reg /CalisMTBol /CalistoMT /CalistoMT-BoldItalic /CalistoMT-Italic /Cambria /Cambria-Bold /Cambria-BoldItalic /Cambria-Italic /CambriaMath /Candara /Candara-Bold /Candara-BoldItalic /Candara-Italic /Carta /CaslonOpenfaceBT-Regular /Castellar /CastellarMT /Centaur /Centaur-Italic /Century /CenturyGothic /CenturyGothic-Bold /CenturyGothic-BoldItalic /CenturyGothic-Italic /CenturySchL-Bold /CenturySchL-BoldItal /CenturySchL-Ital /CenturySchL-Roma /CenturySchoolbook /CenturySchoolbook-Bold /CenturySchoolbook-BoldItalic /CenturySchoolbook-Italic /CGTimes-Bold /CGTimes-BoldItalic /CGTimes-Italic /CGTimes-Regular /CharterBT-Bold /CharterBT-BoldItalic /CharterBT-Italic /CharterBT-Roman /CheltenhamITCbyBT-Bold /CheltenhamITCbyBT-BoldItalic /CheltenhamITCbyBT-Book /CheltenhamITCbyBT-BookItalic /Chiller-Regular /CMB10 /CMBSY10 /CMBSY5 /CMBSY6 /CMBSY7 /CMBSY8 /CMBSY9 /CMBX10 /CMBX12 /CMBX5 /CMBX6 /CMBX7 /CMBX8 /CMBX9 /CMBXSL10 /CMBXTI10 /CMCSC10 /CMCSC8 /CMCSC9 /CMDUNH10 /CMEX10 /CMEX7 /CMEX8 /CMEX9 /CMFF10 /CMFI10 /CMFIB8 /CMINCH /CMITT10 /CMMI10 /CMMI12 /CMMI5 /CMMI6 /CMMI7 /CMMI8 /CMMI9 /CMMIB10 /CMMIB5 /CMMIB6 /CMMIB7 /CMMIB8 /CMMIB9 /CMR10 /CMR12 /CMR17 /CMR5 /CMR6 /CMR7 /CMR8 /CMR9 /CMSL10 /CMSL12 /CMSL8 /CMSL9 /CMSLTT10 /CMSS10 /CMSS12 /CMSS17 /CMSS8 /CMSS9 /CMSSBX10 /CMSSDC10 /CMSSI10 /CMSSI12 /CMSSI17 /CMSSI8 /CMSSI9 /CMSSQ8 /CMSSQI8 /CMSY10 /CMSY5 /CMSY6 /CMSY7 /CMSY8 /CMSY9 /CMTCSC10 /CMTEX10 /CMTEX8 /CMTEX9 /CMTI10 /CMTI12 /CMTI7 /CMTI8 /CMTI9 /CMTT10 /CMTT12 /CMTT8 /CMTT9 /CMU10 /CMVTT10 /ColonnaMT /Colossalis-Bold /ComicSansMS /ComicSansMS-Bold /Consolas /Consolas-Bold /Consolas-BoldItalic /Consolas-Italic /Constantia /Constantia-Bold /Constantia-BoldItalic /Constantia-Italic /CooperBlack /CopperplateGothic-Bold /CopperplateGothic-Light /Copperplate-ThirtyThreeBC /Corbel /Corbel-Bold /Corbel-BoldItalic /Corbel-Italic /CordiaNew /CordiaNew-Bold /CordiaNew-BoldItalic /CordiaNew-Italic /CordiaUPC /CordiaUPC-Bold /CordiaUPC-BoldItalic /CordiaUPC-Italic /Courier /Courier-Bold /Courier-BoldOblique /CourierNewPS-BoldItalicMT /CourierNewPS-BoldMT /CourierNewPS-ItalicMT /CourierNewPSMT /Courier-Oblique /CourierStd /CourierStd-Bold /CourierStd-BoldOblique /CourierStd-Oblique /CourierX-Bold /CourierX-BoldOblique /CourierX-Oblique /CourierX-Regular /CreepyRegular /CurlzMT /David-Bold /David-Reg /DavidTransparent /Desdemona /DilleniaUPC /DilleniaUPCBold /DilleniaUPCBoldItalic /DilleniaUPCItalic /Dingbats /DomCasual /Dotum /DotumChe /EdwardianScriptITC /Elephant-Italic /Elephant-Regular /EngraversGothicBT-Regular /EngraversMT /EraserDust /ErasITC-Bold /ErasITC-Demi /ErasITC-Light /ErasITC-Medium /ErieBlackPSMT /ErieLightPSMT /EriePSMT /EstrangeloEdessa /Euclid /Euclid-Bold /Euclid-BoldItalic /EuclidExtra /EuclidExtra-Bold /EuclidFraktur /EuclidFraktur-Bold /Euclid-Italic /EuclidMathOne /EuclidMathOne-Bold /EuclidMathTwo /EuclidMathTwo-Bold /EuclidSymbol /EuclidSymbol-Bold /EuclidSymbol-BoldItalic /EuclidSymbol-Italic /EucrosiaUPC /EucrosiaUPCBold /EucrosiaUPCBoldItalic /EucrosiaUPCItalic /EUEX10 /EUEX7 /EUEX8 /EUEX9 /EUFB10 /EUFB5 /EUFB7 /EUFM10 /EUFM5 /EUFM7 /EURB10 /EURB5 /EURB7 /EURM10 /EURM5 /EURM7 /EuroMono-Bold /EuroMono-BoldItalic /EuroMono-Italic /EuroMono-Regular /EuroSans-Bold /EuroSans-BoldItalic /EuroSans-Italic /EuroSans-Regular /EuroSerif-Bold /EuroSerif-BoldItalic /EuroSerif-Italic /EuroSerif-Regular /EuroSig /EUSB10 /EUSB5 /EUSB7 /EUSM10 /EUSM5 /EUSM7 /FelixTitlingMT /Fences /FencesPlain /FigaroMT /FixedMiriamTransparent /FootlightMTLight /Formata-Italic /Formata-Medium /Formata-MediumItalic /Formata-Regular /ForteMT /FranklinGothic-Book /FranklinGothic-BookItalic /FranklinGothic-Demi /FranklinGothic-DemiCond /FranklinGothic-DemiItalic /FranklinGothic-Heavy /FranklinGothic-HeavyItalic /FranklinGothicITCbyBT-Book /FranklinGothicITCbyBT-BookItal /FranklinGothicITCbyBT-Demi /FranklinGothicITCbyBT-DemiItal /FranklinGothic-Medium /FranklinGothic-MediumCond /FranklinGothic-MediumItalic /FrankRuehl /FreesiaUPC /FreesiaUPCBold /FreesiaUPCBoldItalic /FreesiaUPCItalic /FreestyleScript-Regular /FrenchScriptMT /Frutiger-Black /Frutiger-BlackCn /Frutiger-BlackItalic /Frutiger-Bold /Frutiger-BoldCn /Frutiger-BoldItalic /Frutiger-Cn /Frutiger-ExtraBlackCn /Frutiger-Italic /Frutiger-Light /Frutiger-LightCn /Frutiger-LightItalic /Frutiger-Roman /Frutiger-UltraBlack /Futura-Bold /Futura-BoldOblique /Futura-Book /Futura-BookOblique /FuturaBT-Bold /FuturaBT-BoldItalic /FuturaBT-Book /FuturaBT-BookItalic /FuturaBT-Medium /FuturaBT-MediumItalic /Futura-Light /Futura-LightOblique /GalliardITCbyBT-Bold /GalliardITCbyBT-BoldItalic /GalliardITCbyBT-Italic /GalliardITCbyBT-Roman /Garamond /Garamond-Bold /Garamond-BoldCondensed /Garamond-BoldCondensedItalic /Garamond-BoldItalic /Garamond-BookCondensed /Garamond-BookCondensedItalic /Garamond-Italic /Garamond-LightCondensed /Garamond-LightCondensedItalic /Gautami /GeometricSlab703BT-Light /GeometricSlab703BT-LightItalic /Georgia /Georgia-Bold /Georgia-BoldItalic /Georgia-Italic /GeorgiaRef /Giddyup /Giddyup-Thangs /Gigi-Regular /GillSans /GillSans-Bold /GillSans-BoldItalic /GillSans-Condensed /GillSans-CondensedBold /GillSans-Italic /GillSans-Light /GillSans-LightItalic /GillSansMT /GillSansMT-Bold /GillSansMT-BoldItalic /GillSansMT-Condensed /GillSansMT-ExtraCondensedBold /GillSansMT-Italic /GillSans-UltraBold /GillSans-UltraBoldCondensed /GloucesterMT-ExtraCondensed /Gothic-Thirteen /GoudyOldStyleBT-Bold /GoudyOldStyleBT-BoldItalic /GoudyOldStyleBT-Italic /GoudyOldStyleBT-Roman /GoudyOldStyleT-Bold /GoudyOldStyleT-Italic /GoudyOldStyleT-Regular /GoudyStout /GoudyTextMT-LombardicCapitals /GSIDefaultSymbols /Gulim /GulimChe /Gungsuh /GungsuhChe /Haettenschweiler /HarlowSolid /Harrington /Helvetica /Helvetica-Black /Helvetica-BlackOblique /Helvetica-Bold /Helvetica-BoldOblique /Helvetica-Condensed /Helvetica-Condensed-Black /Helvetica-Condensed-BlackObl /Helvetica-Condensed-Bold /Helvetica-Condensed-BoldObl /Helvetica-Condensed-Light /Helvetica-Condensed-LightObl /Helvetica-Condensed-Oblique /Helvetica-Fraction /Helvetica-Narrow /Helvetica-Narrow-Bold /Helvetica-Narrow-BoldOblique /Helvetica-Narrow-Oblique /Helvetica-Oblique /HighTowerText-Italic /HighTowerText-Reg /Humanist521BT-BoldCondensed /Humanist521BT-Light /Humanist521BT-LightItalic /Humanist521BT-RomanCondensed /Imago-ExtraBold /Impact /ImprintMT-Shadow /InformalRoman-Regular /IrisUPC /IrisUPCBold /IrisUPCBoldItalic /IrisUPCItalic /Ironwood /ItcEras-Medium /ItcKabel-Bold /ItcKabel-Book /ItcKabel-Demi /ItcKabel-Medium /ItcKabel-Ultra /JasmineUPC /JasmineUPC-Bold /JasmineUPC-BoldItalic /JasmineUPC-Italic /JoannaMT /JoannaMT-Italic /Jokerman-Regular /JuiceITC-Regular /Kartika /Kaufmann /KaufmannBT-Bold /KaufmannBT-Regular /KidTYPEPaint /KinoMT /KodchiangUPC /KodchiangUPC-Bold /KodchiangUPC-BoldItalic /KodchiangUPC-Italic /KorinnaITCbyBT-Regular /KozGoProVI-Medium /KozMinProVI-Regular /KristenITC-Regular /KunstlerScript /Latha /LatinWide /LetterGothic /LetterGothic-Bold /LetterGothic-BoldOblique /LetterGothic-BoldSlanted /LetterGothicMT /LetterGothicMT-Bold /LetterGothicMT-BoldOblique /LetterGothicMT-Oblique /LetterGothic-Slanted /LetterGothicStd /LetterGothicStd-Bold /LetterGothicStd-BoldSlanted /LetterGothicStd-Slanted /LevenimMT /LevenimMTBold /LilyUPC /LilyUPCBold /LilyUPCBoldItalic /LilyUPCItalic /Lithos-Black /Lithos-Regular /LotusWPBox-Roman /LotusWPIcon-Roman /LotusWPIntA-Roman /LotusWPIntB-Roman /LotusWPType-Roman /LucidaBright /LucidaBright-Demi /LucidaBright-DemiItalic /LucidaBright-Italic /LucidaCalligraphy-Italic /LucidaConsole /LucidaFax /LucidaFax-Demi /LucidaFax-DemiItalic /LucidaFax-Italic /LucidaHandwriting-Italic /LucidaSans /LucidaSans-Demi /LucidaSans-DemiItalic /LucidaSans-Italic /LucidaSans-Typewriter /LucidaSans-TypewriterBold /LucidaSans-TypewriterBoldOblique /LucidaSans-TypewriterOblique /LucidaSansUnicode /Lydian /Magneto-Bold /MaiandraGD-Regular /Mangal-Regular /Map-Symbols /MathA /MathB /MathC /Mathematica1 /Mathematica1-Bold /Mathematica1Mono /Mathematica1Mono-Bold /Mathematica2 /Mathematica2-Bold /Mathematica2Mono /Mathematica2Mono-Bold /Mathematica3 /Mathematica3-Bold /Mathematica3Mono /Mathematica3Mono-Bold /Mathematica4 /Mathematica4-Bold /Mathematica4Mono /Mathematica4Mono-Bold /Mathematica5 /Mathematica5-Bold /Mathematica5Mono /Mathematica5Mono-Bold /Mathematica6 /Mathematica6Bold /Mathematica6Mono /Mathematica6MonoBold /Mathematica7 /Mathematica7Bold /Mathematica7Mono /Mathematica7MonoBold /MatisseITC-Regular /MaturaMTScriptCapitals /Mesquite /Mezz-Black /Mezz-Regular /MICR /MicrosoftSansSerif /MingLiU /Minion-BoldCondensed /Minion-BoldCondensedItalic /Minion-Condensed /Minion-CondensedItalic /Minion-Ornaments /MinionPro-Bold /MinionPro-BoldIt /MinionPro-It /MinionPro-Regular /MinionPro-Semibold /MinionPro-SemiboldIt /Miriam /MiriamFixed /MiriamTransparent /Mistral /Modern-Regular /MonotypeCorsiva /MonotypeSorts /MSAM10 /MSAM5 /MSAM6 /MSAM7 /MSAM8 /MSAM9 /MSBM10 /MSBM5 /MSBM6 /MSBM7 /MSBM8 /MSBM9 /MS-Gothic /MSHei /MSLineDrawPSMT /MS-Mincho /MSOutlook /MS-PGothic /MS-PMincho /MSReference1 /MSReference2 /MSReferenceSansSerif /MSReferenceSansSerif-Bold /MSReferenceSansSerif-BoldItalic /MSReferenceSansSerif-Italic /MSReferenceSerif /MSReferenceSerif-Bold /MSReferenceSerif-BoldItalic /MSReferenceSerif-Italic /MSReferenceSpecialty /MSSong /MS-UIGothic /MT-Extra /MT-Symbol /MT-Symbol-Italic /MVBoli /Myriad-Bold /Myriad-BoldItalic /Myriad-Italic /MyriadPro-Black /MyriadPro-BlackIt /MyriadPro-Bold /MyriadPro-BoldIt /MyriadPro-It /MyriadPro-Light /MyriadPro-LightIt /MyriadPro-Regular /MyriadPro-Semibold /MyriadPro-SemiboldIt /Myriad-Roman /Narkisim /NewCenturySchlbk-Bold /NewCenturySchlbk-BoldItalic /NewCenturySchlbk-Italic /NewCenturySchlbk-Roman /NewMilleniumSchlbk-BoldItalicSH /NewsGothic /NewsGothic-Bold /NewsGothicBT-Bold /NewsGothicBT-BoldItalic /NewsGothicBT-Italic /NewsGothicBT-Roman /NewsGothic-Condensed /NewsGothic-Italic /NewsGothicMT /NewsGothicMT-Bold /NewsGothicMT-Italic /NiagaraEngraved-Reg /NiagaraSolid-Reg /NimbusMonL-Bold /NimbusMonL-BoldObli /NimbusMonL-Regu /NimbusMonL-ReguObli /NimbusRomNo9L-Medi /NimbusRomNo9L-MediItal /NimbusRomNo9L-Regu /NimbusRomNo9L-ReguItal /NimbusSanL-Bold /NimbusSanL-BoldCond /NimbusSanL-BoldCondItal /NimbusSanL-BoldItal /NimbusSanL-Regu /NimbusSanL-ReguCond /NimbusSanL-ReguCondItal /NimbusSanL-ReguItal /Nimrod /Nimrod-Bold /Nimrod-BoldItalic /Nimrod-Italic /NSimSun /Nueva-BoldExtended /Nueva-BoldExtendedItalic /Nueva-Italic /Nueva-Roman /NuptialScript /OCRA /OCRA-Alternate /OCRAExtended /OCRB /OCRB-Alternate /OfficinaSans-Bold /OfficinaSans-BoldItalic /OfficinaSans-Book /OfficinaSans-BookItalic /OfficinaSerif-Bold /OfficinaSerif-BoldItalic /OfficinaSerif-Book /OfficinaSerif-BookItalic /OldEnglishTextMT /Onyx /OnyxBT-Regular /OzHandicraftBT-Roman /PalaceScriptMT /Palatino-Bold /Palatino-BoldItalic /Palatino-Italic /PalatinoLinotype-Bold /PalatinoLinotype-BoldItalic /PalatinoLinotype-Italic /PalatinoLinotype-Roman /Palatino-Roman /PapyrusPlain /Papyrus-Regular /Parchment-Regular /Parisian /ParkAvenue /Penumbra-SemiboldFlare /Penumbra-SemiboldSans /Penumbra-SemiboldSerif /PepitaMT /Perpetua /Perpetua-Bold /Perpetua-BoldItalic /Perpetua-Italic /PerpetuaTitlingMT-Bold /PerpetuaTitlingMT-Light /PhotinaCasualBlack /Playbill /PMingLiU /Poetica-SuppOrnaments /PoorRichard-Regular /PopplLaudatio-Italic /PopplLaudatio-Medium /PopplLaudatio-MediumItalic /PopplLaudatio-Regular /PrestigeElite /Pristina-Regular /PTBarnumBT-Regular /Raavi /RageItalic /Ravie /RefSpecialty /Ribbon131BT-Bold /Rockwell /Rockwell-Bold /Rockwell-BoldItalic /Rockwell-Condensed /Rockwell-CondensedBold /Rockwell-ExtraBold /Rockwell-Italic /Rockwell-Light /Rockwell-LightItalic /Rod /RodTransparent /RunicMT-Condensed /Sanvito-Light /Sanvito-Roman /ScriptC /ScriptMTBold /SegoeUI /SegoeUI-Bold /SegoeUI-BoldItalic /SegoeUI-Italic /Serpentine-BoldOblique /ShelleyVolanteBT-Regular /ShowcardGothic-Reg /Shruti /SimHei /SimSun /SnapITC-Regular /StandardSymL /Stencil /StoneSans /StoneSans-Bold /StoneSans-BoldItalic /StoneSans-Italic /StoneSans-Semibold /StoneSans-SemiboldItalic /Stop /Swiss721BT-BlackExtended /Sylfaen /Symbol /SymbolMT /Tahoma /Tahoma-Bold /Tci1 /Tci1Bold /Tci1BoldItalic /Tci1Italic /Tci2 /Tci2Bold /Tci2BoldItalic /Tci2Italic /Tci3 /Tci3Bold /Tci3BoldItalic /Tci3Italic /Tci4 /Tci4Bold /Tci4BoldItalic /Tci4Italic /TechnicalItalic /TechnicalPlain /Tekton /Tekton-Bold /TektonMM /Tempo-HeavyCondensed /Tempo-HeavyCondensedItalic /TempusSansITC /Times-Bold /Times-BoldItalic /Times-BoldItalicOsF /Times-BoldSC /Times-ExtraBold /Times-Italic /Times-ItalicOsF /TimesNewRomanMT-ExtraBold /TimesNewRomanPS-BoldItalicMT /TimesNewRomanPS-BoldMT /TimesNewRomanPS-ItalicMT /TimesNewRomanPSMT /Times-Roman /Times-RomanSC /Trajan-Bold /Trebuchet-BoldItalic /TrebuchetMS /TrebuchetMS-Bold /TrebuchetMS-Italic /Tunga-Regular /TwCenMT-Bold /TwCenMT-BoldItalic /TwCenMT-Condensed /TwCenMT-CondensedBold /TwCenMT-CondensedExtraBold /TwCenMT-CondensedMedium /TwCenMT-Italic /TwCenMT-Regular /Univers-Bold /Univers-BoldItalic /UniversCondensed-Bold /UniversCondensed-BoldItalic /UniversCondensed-Medium /UniversCondensed-MediumItalic /Univers-Medium /Univers-MediumItalic /URWBookmanL-DemiBold /URWBookmanL-DemiBoldItal /URWBookmanL-Ligh /URWBookmanL-LighItal /URWChanceryL-MediItal /URWGothicL-Book /URWGothicL-BookObli /URWGothicL-Demi /URWGothicL-DemiObli /URWPalladioL-Bold /URWPalladioL-BoldItal /URWPalladioL-Ital /URWPalladioL-Roma /USPSBarCode /VAGRounded-Black /VAGRounded-Bold /VAGRounded-Light /VAGRounded-Thin /Verdana /Verdana-Bold /Verdana-BoldItalic /Verdana-Italic /VerdanaRef /VinerHandITC /Viva-BoldExtraExtended /Vivaldii /Viva-LightCondensed /Viva-Regular /VladimirScript /Vrinda /Webdings /Westminster /Willow /Wingdings2 /Wingdings3 /Wingdings-Regular /WNCYB10 /WNCYI10 /WNCYR10 /WNCYSC10 /WNCYSS10 /WoodtypeOrnaments-One /WoodtypeOrnaments-Two /WP-ArabicScriptSihafa /WP-ArabicSihafa /WP-BoxDrawing /WP-CyrillicA /WP-CyrillicB /WP-GreekCentury /WP-GreekCourier /WP-GreekHelve /WP-HebrewDavid /WP-IconicSymbolsA /WP-IconicSymbolsB /WP-Japanese /WP-MathA /WP-MathB /WP-MathExtendedA /WP-MathExtendedB /WP-MultinationalAHelve /WP-MultinationalARoman /WP-MultinationalBCourier /WP-MultinationalBHelve /WP-MultinationalBRoman /WP-MultinationalCourier /WP-Phonetic /WPTypographicSymbols /XYATIP10 /XYBSQL10 /XYBTIP10 /XYCIRC10 /XYCMAT10 /XYCMBT10 /XYDASH10 /XYEUAT10 /XYEUBT10 /ZapfChancery-MediumItalic /ZapfDingbats /ZapfHumanist601BT-Bold /ZapfHumanist601BT-BoldItalic /ZapfHumanist601BT-Demi /ZapfHumanist601BT-DemiItalic /ZapfHumanist601BT-Italic /ZapfHumanist601BT-Roman /ZWAdobeF ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 200 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 300 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 2.00333 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >>
/ColorImageDict << /QFactor 1.30 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >>
/JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 10 >>
/JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 10 >>
/AntiAliasGrayImages false
/CropGrayImages true
/GrayImageMinResolution 200
/GrayImageMinResolutionPolicy /OK
/DownsampleGrayImages true
/GrayImageDownsampleType /Bicubic
/GrayImageResolution 300
/GrayImageDepth -1
/GrayImageMinDownsampleDepth 2
/GrayImageDownsampleThreshold 2.00333
/EncodeGrayImages true
/GrayImageFilter /DCTEncode
/AutoFilterGrayImages true
/GrayImageAutoFilterStrategy /JPEG
/GrayACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >>
/GrayImageDict << /QFactor 1.30 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >>
/JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 10 >>
/JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 10 >>
/AntiAliasMonoImages false
/CropMonoImages true
/MonoImageMinResolution 400
/MonoImageMinResolutionPolicy /OK
/DownsampleMonoImages true
/MonoImageDownsampleType /Bicubic
/MonoImageResolution 600
/MonoImageDepth -1
/MonoImageDownsampleThreshold 1.00167
/EncodeMonoImages true
/MonoImageFilter /CCITTFaxEncode
/MonoImageDict << /K -1 >>
/AllowPSXObjects false
/CheckCompliance [
/None
]
/PDFX1aCheck false
/PDFX3Check false
/PDFXCompliantPDFOnly false
/PDFXNoTrimBoxError true
/PDFXTrimBoxToMediaBoxOffset [
0.00000
0.00000
0.00000
0.00000
]
/PDFXSetBleedBoxToMediaBox true
/PDFXBleedBoxToTrimBoxOffset [
0.00000
0.00000
0.00000
0.00000
]
/PDFXOutputIntentProfile (None)
/PDFXOutputConditionIdentifier ()
/PDFXOutputCondition ()
/PDFXRegistryName ()
/PDFXTrapped /False

/CreateJDFFile false
/Description << /ARA
/BGR
/CHS
/CHT
/CZE
/DAN
/DEU
/ESP
/ETI
/FRA
/GRE
/HEB
/HRV
/HUN
/ITA
/JPN
/KOR
/LTH
/LVI
/NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken die zijn geoptimaliseerd voor weergave op een beeldscherm, e-mail en internet. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.)
/NOR
/POL
/PTB
/RUM
/RUS
/SKY
/SLV
/SUO
/SVE
/TUR
/UKR
/ENU (Use these settings to create Adobe PDF documents best suited for on-screen display, e-mail, and the Internet. Created PDF documents can be opened with Acrobat and Adobe Reader 5.0 and later.)
>>
/Namespace [
(Adobe)
(Common)
(1.0)
]
/OtherNamespaces [
<< /AsReaderSpreads false /CropImagesToFrames true /ErrorControl /WarnAndContinue /FlattenerIgnoreSpreadOverrides false /IncludeGuidesGrids false /IncludeNonPrinting false /IncludeSlug false /Namespace [ (Adobe) (InDesign) (4.0) ] /OmitPlacedBitmaps false /OmitPlacedEPS false /OmitPlacedPDF false /SimulateOverprint /Legacy >>
<< /AddBleedMarks false /AddColorBars false /AddCropMarks false /AddPageInfo false /AddRegMarks false /ConvertColors /ConvertToRGB /DestinationProfileName (sRGB IEC61966-2.1) /DestinationProfileSelector /UseName /Downsample16BitImages true /FlattenerPreset << /PresetSelector /MediumResolution >>
/FormElements false
/GenerateStructure false
/IncludeBookmarks false
/IncludeHyperlinks false
/IncludeInteractive false
/IncludeLayers false
/IncludeProfiles true
/MultimediaHandling /UseObjectSettings
/Namespace [
(Adobe)
(CreativeSuite)
(2.0)
]
/PDFXOutputIntentProfileSelector /NA
/PreserveEditing false
/UntaggedCMYKHandling /UseDocumentProfile
/UntaggedRGBHandling /UseDocumentProfile
/UseDocumentBleed false
>>
]
>> setdistillerparams
<< /HWResolution [600 600] /PageSize [612.000 792.000] >> setpagedevice