Weights Extraction And Visualization Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Abstract- The digital information is expanding with exponential curve. It becomes difficult to find the relevant information. In scientific domain, one of the best possibilities to find highly relevant paper is by finding its citations. However, papers receive 100's and sometimes 1000's of citations. With this abdunt information, it becomes difficult to find the highly relevant paper from citations. Sometimes, the citations are just made to give a background study of the field and there is no strong relationship between cited and cited by article. In the current work, we have proposed a system to rank the top 10 cited papers for a cited by paper. The system employs number of weights to find the most relevant paper such as topic weight, keyword weight, and author weight. The manual inspection has revealed that the top ranked papers were actually the most relevant papers.

Keywords- Visualization, Citation, Weight,


The scientific literature is growing exponentially, numbers of new ideas are publishing on the daily basis, and these works is based on the some existing work. The majority of fresh generated ideas are impressed by previously published work. Citation is an important indicator for the quality of a paper and also the quality of researchers. Citation is also defining the quality of journals. Measurements of these different factors are based on finding all relevant citations. This problem boosted the consequence of citation mining tools. Numbers of tools are exists for example ISI Web of Knowledge, Google Scholar, and CiteSeer.

To find the relevant information in the number of publish documents are very difficult in the scientific community. To overcome that issue one of the best possibility is find the citation of the paper. But this has different evils which are highlighted later.

Visualization is defined as "visualization is the process of interpreting the things into visible visual terms or putting into visible form [4]. Information Visualization is defined as "The process of transformation data, information, and knowledge into visual form making use of human's natural capabilities." Information visualization is most critical field of the different business and education domains. We present a new innovative idea of visualization of papers and its citation and weights of these.

In scientific literature relevant papers are finds through citation of papers. In the domain of scientific literature particular one paper could be receive thousand of citations but all these citation are not exactly relevant to the cited paper. Sometime these citation are just for complete the background study of paper. There are no muscular associations between cited and cited by paper. Another important problem is citation is important factor to define the experts of the systems and also to define the paper contribution. To solve these problems we introduce the innovative idea, in which find the relationship between the paper and its citation. For this use the weighting mechanism. The weighting mechanism is based on three weights 1) keyword weight 2) category Weight 3) Common Author Weight.

For checking the performance of the proposed idea we perform the experiments on the dataset of journal of universal of computer science (JUCS). JUCS database is based on the more than 3000 of author's records and 6090 records of papers. Proposed technique gives results which helpful to solve the existing problems of finding the authors relationship with each other and how much strong it. Proposed technique identified the different cluster of authors which helpful in different prospectus of user.

The rest of the organized as follows: Section II gives the some existing techniques overview. Section III gives the detail overview of architecture of proposed technique. Section IV provides the details of proposed system. Section V gives the experimental results and Section VI gives the conclusion of the paper.


Blobby clustering mechanism is Introduce the similarity quantification mathematical framework by using Euclidean vector space [2]. Clustering idea is called blobs. The technique is based on 3d clustering. Blob function to represent particular object and clustering shapes. Different Spatial speckled object locations are defined xi,yi,zi and weight bi features an supplementary scale of sovereignty. Individual Sets of clusters under the boundaries are disjointness and non junction of cluster. The process of cclustering algorithm is, they starts from the center of a object, and after discretization of the field function it searches the nearest volume cell intersecting the isosurface defined by ciso. In the case of isosurface sector of this voxel is previously computed, then the segment of surface are exists and the object is added to the related cluster. In other case isosurface is not computed previously then it is initialized and computed recursively with the cell as a seeding point.

Andrea et al present the Information Aesthetics in Information Visualization model [1]. Aesthetics is the method which ensures the displays remain modest in the bodily settings in which they are positioned. Aesthetics use to refer to the degree of extant on the creative influence on the technique of visualization. Information Aesthetics is defined as an aesthetics measure of quantitative in perspective to the information contented of an image's ingredient parts. Information aesthetics is analysed from an information visualization perspective. There are two factors are define in the model: one of them is data focus and other is mapping technique. Data focus is the gritty by examine how the visualization smooth the progress of knowledge acquirement Mapping technique is a concept which describes the methods employed by a visualization creator to represent an abstract dataset. Mapping techniques are direct vs interpretive. Direct mapping focus is driven by standards learnt from visual cognition. Interpretive mapping involve subjective decision and style influence. Proposed model is facilitating both types of users. Proposed information Aesthetic visualization increase the credibility of resulting visual artifact and user investigate message enforcing data patterns

The list of basic tasks of graph Visualization is defined by Bongshin Lee et al [3]. They suggest the lists of tasks which are encountered while analysing graph data. They exhibit how all difficult duties could be seen as a series of low-level tasks performed on those objects. Discusses the Graph exact Objects, Low Level Tasks and Graph Task Taxonomy. The Graph Specific Objects are nodes, links, paths, graphs, clusters connected components and groups. These objects basically define the structure of the Graph. These basic Graph Specific Objects are used for applying Low-Level Tasks. The Low-Level Tasks is applied on datasets and an aggregation function that generate a numeric demonstration for a set data cases. They also describe the basic tasks which are present in the literature and finally proposed one graph related and two general tasks. In Graph Task Taxonomy they abridge a list of tasks generally encounter while analysing graph data. The errands are topology-based tasks, attribute-based tasks, browsing tasks, and the overview task. These jobs have common descriptions. In addition, they also show how each task can be decomposed into low-level tasks. The topology-based tasks deal with adjacency, accessibility, common connection and connectivity. The attribute-based tasks deal with the count tasks on the link or nodes attributes. The Browsing task deals with getting the estimated values. There are also High Level Tasks which are not covered by the low level tasks defined in the technique. For Example how graph change over time.

The TreePlus visualization Lee et al display graph data as a least spanning tree classify to sustain readability labels and ease layout [4]. TreePlus track the plant a seed and observe it grow up design method, in which a node in the graph is elected to be the root of the tree and a spanning tree of the graph from that node is construct and display. The tree-based layout allocate only for a restricted summary of data. On the other hand, the design of TreePlus allocate users to quickly read label and analyse the significance of relationships, that construct to a very muscular tool to explore local areas of a graph in detail. In a chore which include discovered a specific node with a maximum number of connections to another type of node, users preferred an orderly browsing. These methods enable user to interactively investigate a graph by initial at a node and then expand incrementally and explore the graph. They transform the graph into the tree extracting minimum spanning tree. If the graph don't have any root node then 1) the node that has the most links and 2) the node whose cumulative distance to all other nodes is minimal.

Jia et al proposed the idea of filtering edges with small between's centrality [5]. Deterministic sieve use, as its name recommends, a deterministic algorithm for selecting of the nodes/edges to be removed. This filtering mechanism based on node/edge attribute on topologic value such as between centrality, or other graph property. For example, filter based on edge-between centrality are used to exclusion of less important edges while keep the underlying organization of the graph.

MoireGraphs merge new focus and context radial graph layout with a group of interaction technique like focus strength changing, radial rotation, level highlighting, animated transitions and node information to assist in the exploration of graphs with visual nodes [6]. Three steps of graph construction is: a focus is chosen, tree is located by means of the fresh radial focus+context layout method, and the tree its visual node elements are rendered.

JUCS Database




Assign Weights

Paper id





Y Ed tool

Keyword weight

Number of keyword matched in paper and cited by =keyword weight else keyword weight is equal to 0

Category weight

If paper and cited paper have same category then assign 1 otherwise 0

Author weight

If paper and citied paper have common author then author weight is1and no common author then assign 0.

Figure 1 Architecture of citation weighty mechanism.

proposed Architect

The proposed architecture is show in the figure.1. This figure shows the different modules of the proposed architecture and work flow of the proposed system. As we show in the figure first of all extracting the papers and citation of theses extracted papers. The next module is show three different works, extracting the keyword form both of the paper. Secondly find the category of the paper and finally find the authors of these two papers. Next module shows the weights mechanism of the system. In this module as shown three sub modules are defined, these are author weight, keyword weight and third is category weight. Author weight module is worked

as if paper and citied paper have common author then author weight is 0 otherwise 1. As it is keyword weight is work as if number of keyword matched in paper and cited by =keyword weight else keyword weight is equal to 0. Finally the last module of weighting mechanism is category weight. The category weight is defined as if paper and cited paper have same category then assign 1 otherwise 0. After finding the weight the next step is assign the weights. Finally visualize these weights by using the Y.Ed tool.

Citation relationship mechanism

Proposed the innovative idea, in which find the relationship between the paper and its citation and another contribution is visualization. We use the weighting mechanism for finding the relationship between paper and its citation. The weighting mechanism is: we define the three weights based on different views. These weights are:

1) Keyword weight

2) Category Weight

3) Common Author Weight.

Procedure of Weighting Mechanism:

Keyword weight:

Keyword weight is assign as if both papers have the same keyword (more than one) assign 1 and if no keyword match assigns 0. Otherwise total keyword matched/ cited paper keyword.

Category Weight:

Category weight is import. If both papers have the same category then assign 1 otherwise it assign is 0.

Common Author weight:

Common author weight is assigned as: if both papers have the same author than assign 1 and if both paper have no common paper assign the 0 weight. Otherwise total common author /cited paper author

Figure 1


We use the dataset Journal of Universal Computer Science (JUCS). Extract the authors and its publication in this datasets. JUCS dataset contain the 6090 distinct number of publications and 11003 distinct authors data.


We use the yEd is a powerful tool. Y.Ed is used to quickly and effectively generate high-quality diagrams. In yEd we can create your diagrams manually or import your external data. Y.Ed is freely available and runs on all major operating systems.

Experimental results.

To assess the performance of proposed framework for assign the weights of citations, we use the Journal of Universal of Computers (JUCS) database. Perform experiments on 6090 publications records, and 11003 author's records. Proposed framework is implemented in the with java, integrated development environment java beans and it performed on a Intel PC with 1GB of main memory and 3 GHz processor.

Table 1 show the final results of the proposed system. In this table show that the paper id and cited by paper id and their weights.

Paper ID

Cited By ID
































Table 1

Figure 2 represent the weights graphs. Horizontal Axis label show the paper id and vertical axis show the weights.

case study

In the following table gives the information about the paper id, citedby paper id, title of the paper, title of the cited by paper, keywords of the both papers cited and citedby paper, both papers authors , category of cited and citedby paper and extracted weights. Match keywords, authors, categories are highlighted in the red front. For example paperid 252 and cited by id 250 have weight 8 because all keyword are matched with the cited paper and both paper have lie in the same categories and also have common authors then weight of the paper is maximum and it is highly relevant to the cited paper. Same as 486 and 484 assign the weight 3 because no author match, only 2 keyword are match and paper lie in the


Cited by id





Cited By Title

Cited by Keyword


Citedby category




AspectLua - A Dynamic AOP Approach

AOP ,MOP, Lua , dynamic aspects, reflection



RE-AspectLua - Achieving Reuse in AspectLua

AOP, Lua, dynamic aspects, reusability, heterogeneity






On Implementing EREW Work-Optimally on Mesh of Trees

EREW, mesh of trees, shared memory, simulation, work-optimal, randomized, coated mesh.



Balanced PRAM Simulations via Moving Threads and Hashing

balanced workload, moving threads, EREW, shared memory, simulation, work-optimal, randomized, OCPC, mesh of trees, coated mesh






MONSTR I - Fundamental Issues and the Design of MONSTR

Intermediate Languages, Term Graph Rewriting, MONSTR, Semantic Models



164, 170

MONSTR II - Suspending Semantics and Independence

Intermediate Languages, Term Graph Rewriting, MONSTR, Semantic Models


97,121, 164,170




A Translation of the Pi-Calculus Into MONSTR

Concurrency, Pi-Calculus, Term Graph Rewriting, MONSTR, Process Networks, Simulation, Serialisability.

677, 1517, 1518

97,121,164, 170

MONSTR II - Suspending Semantics and Independence

Intermediate Languages, Term Graph Rewriting, MONSTR, Semantic Models


97,121, 164,170




A Logic for Abstract State Machines

Abstract State Machines, dynamic logic, modal logic, logical foundations of specification languages


108,163 ,169

Verification of ASM Refinements Using Generalized Forward Simulation

refine, ment, forward simulation, data refinement, commuting diagrams, Abstract State Machines, transition systems, I/O-Automata, dynamic logic, correctness proofs, interactive theorem proving, compiler cerification


105,163, 108 , 164,169




SEAL-II - The Soft Spot between Richly Structured and Unstructured Knowledge

knowledge portal, ontology



Managing User Focused Access to Distributed Knowledge

knowledge portal, information integration, ontology






Foundations of MIRACLE: Multimedia Information Repository, A Computer-supported Language Effort

visual languages, visual communication, constructed languages, computer-supported communication, information archiving, information retrieval ,language independent communication, converging technology, multimedia


3,224,230 ,234, 321,322

A Cross-Disciplinary Bibliography on Visual Languages for Information Sharing and Archiving

visual languages, visual communication, constructed languages, sign languages, interactive maps, computer-supported communication, information archiving, information retrieval, language independent communication


4,224, 230, 234,321, 322


same category. So the final weight is 2 because of author weight is 0 and keyword weight is 0.4.


Citation of paper is most important and critical aspect of the scientific literature. Different types of assessments are done through the citation for example quality of a paper, quality of researchers and quality of journals is asses through citation. So these critical importances increase the problem of finding relevant citation and remove the wake citation. We propose the novel approaches for solving this problem for that, we assign the weights of citation on the basis of different weights. Proposed system based on three weights 1) keyword weight 2) category Weight 3) Common Author Weight.