Exploiting Literature-Based Discovery to Study Effects of Bullying S.M.Shamimul Hasan Computational Sciences and Engineering Division Oak Ridge National Laboratory Oak Ridge, TN, USA hasans@ornl.gov ABSTRACT ACM Reference Format: Bullying represents an aggressive behavioral problem. It can S.M.Shamimul Hasan. 2019. Exploiting Literature-Based Discovery to Study Effects of Bullying. In Joint Proceedings of the ACM IUI create severe mental illness in children, adolescents, and 2019 Workshops, Los Angeles, USA, March 20, 2019 , 8 pages. adults. Researchers are interested in studying bullying to understand its long-term negative consequences and design better-targeted intervention programs. Broadly researchers 1 INTRODUCTION in psychology and education have conducted studies on bul- Bullying is an abuse delivered in the form of threatening, lying without using advanced analytical features offered by demeaning, and belittling a person by a physically stronger the computer science domain. However, researchers could individual or one with a higher social status than the vic- benefit from power of advanced analytics for an in-depth tim [18]. Children, adolescents, and even adults are victims understanding of bullying. In this paper, we present graph- of bullying [21]. However, young people are at the high- based advanced analytics by employing a literature-based est risk [1]. Bullying can create serious mental distress [10]. discovery (LBD) approach to study the direct and indirect Moreover, combined with additional risk factors, being in- consequences of bullying from published literature. In addi- volved in bullying, increases the possibility of engaging in tion, we describe a graph-based study on the evolution of re- suicide-related behaviors [10]. Furthermore, not only the search on bullying consequences in the last two decades. Our victims but the atmosphere is affected by bullying. Research graph-based study reveals a summary view of associations studies show that individuals who experienced bullying be- of bullying with other diseases and disorders from published havior problems at school age have a tendency to participate literature. Moreover, we employ literature reference infor- in criminal behavior as adults, including racial attacks, dam- mation to quantify the strength of the graph-derived rela- age to public property, sexual attacks and even killing [8]. tionships. Furthermore, we propose an interactive Graphical Bullying is an increasingly interesting topic for researchers, User Interface (GUI) design to study bullying. We demon- because of growing evidence demonstrating long-term nega- strate that LBD is a valuable approach to perform advanced tive outcomes for both bullies and victims. The researchers analytics on bullying that helps us to understand the severity aim to identify the causes of bullying in order to find effec- of its consequences. tive intervention programs targeting the risk and protective factors related to the involvement in bullying. However, it is CCS CONCEPTS difficult to study bullying behavior because it is frequently • Software and its engineering → Software creation covered up or fabricated (eheck [22] for detail) [36]. Histori- and management. cally, experts in psychology and education have conducted studies on bullying but other fields such as computer science, big data, and virtual communities can provide a significant KEYWORDS contribution by offering new perspectives and innovative bullying, literature-based discovery, graph analytics approaches to studying bullying and developing efficient bullying interventions [15]. Despite the fact that the bully-victim problem had existed previously, the first systematic studies were only conducted in the early 1970s [26, 34]. In the early stage, studies were IUI Workshops’19, March 20, 2019, Los Angeles, USA performed in Scandinavia, US, Japan, Australia, and UK in Copyright © 2019 for the individual papers by the papers’ authors. Copying permitted for private and academic purposes. This volume is published and the, 1970s, 1980s and 1990s [27]. As stated in US national copyrighted by its editors. data, approximately 15% of youngsters experienced cyber- bullying and 20% were bullied at school in 2013. Moreover, IUI Workshops’19, March 20, 2019, Los Angeles, USA Trovato and Tobin, et al. 8% of students reported being bullied daily [15]. The rate of Semantic MEDLINE bullying among children is fairly consistent; nevertheless, Semantic MEDLINE is a web-based application that pro- researchers have found bullying among adults as well [29]. vides biomedical document retrieval, summarization, and This phenomenon is particularly prominent in the work en- visualization services [30]. This application uses MEDLINE vironment. According to the “Workplace Bullying Survey citations generated from PubMed searches [19, 30]. Semantic 2014,” 72% of Americans reported being aware of bullying in MEDLINE employs a natural language processing tool called the workplace, 21% witnessed it, and 27% suffered abusive SemRep to find semantic relationships (or predications) from conduct at work [37]. the titles and abstracts of PubMed citations [38]. SemRep In this paper, we present a summarized view of bully- examines each sentence in PubMed titles and abstracts to ing’s direct and indirect consequences. The literature-based extract meaningful relationships [30]. For example, SemRep discovery (LBD) method is employed to identify bullying’s obtains the relation psychological distress COEXISTS_WITH short-term and long-term impacts from the published litera- bullying from the following citation “The Impact of School ture. Moreover, we present a systematic discussion on how Social Support and Bullying Victimization on Psychologi- research on bullying consequences evolves over time. We cal Distress among California Adolescents (PMID:27708555). utilize Semantic MEDLINE, a web application that manages However, all the SemRep relations are not meaningful. For the outcomes of PubMed searches by visualizing SemRep example, Semantic MEDLINE shows that citation [14] cre- predications collected from MEDLINE citations [24]. We ates bullying and vitelliform dystrophy relationship. However, demonstrate that LBD is a useful approach for domain re- there exists no such relationship in citation [14]. Thus, it is searchers, to discover logical relationships between bullying necessary for us to read the citations carefully to find out and its harmful impacts. the soundness and strength of the relationships. We review previous work on LBD in Section 2. In Sec- SemRep relationships are presented in a triple form as it tion 3, we present our research methodology. In Section 4, contains three parts: subject, predicate, and object. In the we provide a detail discussion of our research results. In Sec- above example, the subject is psychological distress, the pred- tion 5, we discuss propose user interface design and Section 6 icate is COEXISTS_WITH, and the object is bullying. Sub- concludes the paper. jects and objects represent concepts taken from the UMLS’s Metathesaurus, and predicates are collected from the Se- mantic Network (SN) [24]. Semantic MEDLINE provides vi- 2 BACKGROUND sualization of the relationships as a graph (or network) by Literature-based discovery (LBD) uses published literature considering the subject and object as nodes, and the predicate to identify new interesting relationships between existing as an edge. The graph edge provides citation information knowledge regarding a given phenomenon [33]. One of the and the sentence from which the relationship is extracted. early work on LBD was conducted by Don R. Swanson. The We present an example visualization in Fig. 1 where psycho- author employed LBD technique to study fish oil [33]. Swan- logical distress and bullying maintain a COEXISTS_WITH son apply LBD to find the connection between Migraine and relationship (left side of the figure). The citation sentence Magnesium as well [12]. Hristovski, Friedman, Rindflesch, from which the relationship is extracted is shown in red on and Peterlin proposed to use semantic predication to increase the right side of the figure. Semantic MEDLINE provides nu- the strength of the LBD [17]. Miller et al. employed LBD to do merous node types and relations (edges) based on a citation hypothesis testing. In their study, authors found that testos- search. We refer readers to the SemMed documentation for terone and sleep have a connection via cortisol [24]. Gordon additional discussion on this topic [4]. and Lindsay utilized LBD to find the link between Raynaud’s disease and dietary fish oil [13]. Find Effects of Bullying We utilize semantic MEDLINE to discover direct and indirect 3 METHODOLOGY effects of bullying. We use the free UMLS license to access There are two main components of our research. First, we the semantic MEDLINE. study the direct and indirect effects of bullying. Second, we To find the direct effects of bullying, we query seman- examine the evolution of the patterns depicted in bullying tic MEDLINE with the keyword “bullying” and pursue the consequences in published literature. We employ semantic following steps: MEDLINE tool to conduct our research. We present descrip- tion of the semantic MEDLINE and our research components Step 1: We use the date range (01/01/1900 - 06/30/2017) and below. Throughout the paper, Unified Medical Language the search options (Most recent: 500 citations) to System (UMLS) concept names are italicized. return the relevant citations. Exploiting Literature-Based Discovery to Study Effects of Bullying IUI Workshops’19, March 20, 2019, Los Angeles, USA Figure 1: Semantic MEDLINE web application. Step 2: We summarize the citation by using Summary Type: citation title(s) to collect “Cited By” (another citation refers Treatment and Disease and UMLS concept that occur to a relationship citation) information. in the relations the maximum number of times. We Note: “Cited By” information is one of importance mea- also check the “More Relations” checkbox to utilize surement metric. However, it may not be appropriate for all the relations. Summarization provides us rela- some cases. tions in a subject, predicate, and object format. It also presents the citation sentence (from where the 4 DISCUSSION relation is extracted) and PMID. Effects of Bullying Step 3: We visualize the summarized results in a graph where nodes (UMLS concepts) are the subjects and objects, In the following section, we present our discussion on di- while edges are predicates (from the Semantic Net- rect and indirect effects of bullying based on literature-based work). We use the COEXISTS_WITH relation in our discovery. We query semantic MEDLINE with keyword “bul- study. lying” and use the date range from 01/01/1900 to 06/30/2017, Step 4: We obtain the corresponding citations for each rela- which provides the recent citations on bullying. We obtain tion from the graph edge. Finally, we read the citation many relationships from semantic MEDLINE. Furthermore, to identify meaningful relations. we discuss few relations as examples. To find indirect effects of bullying, we first search semantic Direct Effect. We observe that bullying concept appears di- MEDLINE with all the UMLS concepts that co-occur with rectly with the following concepts: 1) suicide, 2) alcohol and “bullying” (example: Fig. 2-Level 2) and follow the four steps other drugs (AOD) use, 3) metabolic syndrome, 4) psychological mentioned earlier. Next, with the results obtained in the first distress, 5) inflammation, 6) family conflict, and 7) victimizing phase, we search semantic MEDLINE again (see example: others. The relationship between bullying and suicide comes Fig. 2-Level 3) and repeat the above four steps. from the citation [32] where authors study “why bullying leads to suicidal thoughts and behaviors.” The study per- Exploring the Evolution of Research on Bullying formed on 340 depressed adolescents and their ages were We aim to understand the evolution of bullying research 13-19 years. In this study, it was observed that controlling over time. We consider COEXISTS_WITH relationship. We for internalizing symptoms, age, overt and reputational bul- observe from semantic MEDLINE that first UMLS concept lying maintains an association with periodic past month COEXISTS_WITH bullying in the year 1995. Next, from 1995, suicide attempts. Moreover, the authors noted that for male we expand our search date by five-year periods leading to victims, victimization leads to risky behavior engagement bullying COEXISTS_WITH graph for the years 1995, 2000, before suicide attempts whereas for female victims, bullying 2005, 2010, 2015, and 2017 (Semantic MEDLINE provides was directly related to suicide attempts [32]. Bullying and data up to the year of 2017). We generate six graphs with AOD use relation comes from citation [16]. In their study [16], numerous relations. To quantify the importance of a rela- the authors found that in spite of the severity of internalizing tionship, we search Google Scholar [2] with relationship’s problems, substance use maintains independent connections IUI Workshops’19, March 20, 2019, Los Angeles, USA Trovato and Tobin, et al. Figure 2: In this figure, we present our three-level approach example for bullying’s indirect effects identification. In Level 1, we present all the concepts co-occur (first neighbor) with bullying. One of them is psychological distress. In Level 2, we show psychological distress’s first neighbors and Epilepsy is one of them. In Level 3, we display Epilepsy’s first neighbors. with both negative peer norms and bullying. The citation [23] distress has significant impact on epilepsy. We search seman- provides bullying’s co-occurrence with metabolic syndrome tic MEDLINE with epilepsy and find that cognitive deficit has and inflammation. The citation [23] describes that childhood a relationship with epilepsy [31]. bullying has no relationship with metabolic syndrome and We discussed the connection between the bullying and sui- inflammation. Hence, it means that not all the relationships cide before. Now, we search semantic MEDLINE with concept present in semantic MEDLINE are positively correlated. The suicide and find that suicide has relationship with eating dis- relationship between bullying is directly connected to psycho- orders [7]. In the citation [7], via self-image-based prediction, logical distress, and bullied young people are susceptible to the authors showed that suicide has association with eat- experience serious psychological distress (SPD) twice more ing disorders. Subsequently, we search semantic MEDLINE than non-bullied young people [39]. Bullying is also con- with eating disorders and find that it has a relationship with nected to family conflict via psychological distress [35]. Many obesity [25]. In the citation [25], the authors demonstrate a citations (e.g., [28], [9]) in semantic MEDLINE suggest that connection between eating disorders and obesity by distin- bullying concept maintains ISA (sameAS) relationship with guishing “Fear of Obesity” and “Self-Esteem Based on Shape victimizing others. Hence, in our study, we consider them and Weight.” However, their study has some limitations. synonymous. Indirect Effect. In the earlier section (Direct Effect), we men- tion that bullying has a relation with psychological distress [39]. We search semantic MEDLINE with psychological distress and Evolution of Bullying Research Literature find that it has a relationship with at risk for suicide, which Our study shows that bullying research is continuously grow- is shown in citation [6]. In citation [6], the presented results ing as indicated in Fig. 3, which shows relationship of bully- showed that Suicidal behavior is positively associated to psy- ing with other concepts that have been established within chological distress. Next, we search semantic MEDLINE with various time frames. According to Fig. 3(a), aggressive behav- at risk for suicide. We find that at risk for suicide maintains ior is the first disorder that COEXISTS_WITH with bullying relationship with diabetes mellitus suggested by citation [11]. in the year 1995. The 1995 graph contains only two nodes In the citation [11], the authors observe that diabetes mellitus and one edge. After five years, in the year 2000, in addition is strongly connected with an increase in suicidal behaviors to aggressive behavior, bullying is co-occurring with psychi- and ideation. atric problem, victimizing other and social fear (Fig. 3(b)). While bullying has relation with psychological distress [39], The 2000 citation graph contains five nodes and three edges. psychological distress has a relationship with epilepsy [20]. That is, the research work in the time-frame 1995-2000 has In the citation [20], the authors discussed that psychological established these additional relationships. Exploiting Literature-Based Discovery to Study Effects of Bullying IUI Workshops’19, March 20, 2019, Los Angeles, USA (a) Bullying research graph from 01/01/1900 to 06/30/1995. (b) Bullying research graph from 01/01/1900 to 06/30/2000. (c) Bullying research graph from 01/01/1900 to 06/30/2005. (d) Bullying research graph from 01/01/1900 to 06/30/2010. (e) Bullying research graph from 01/01/1900 to 06/30/2015. (f) Bullying research graph from 01/01/1900 to 06/30/2017. Figure 3: In this figure we show, evolution of bullying research literature graph. Here we present all the COEXISTS_WITH relationships. IUI Workshops’19, March 20, 2019, Los Angeles, USA Trovato and Tobin, et al. Considering bullying and victimizing other synonymous, bullying? An exploration of online harassment by known as shown in Fig. 3(c), in the year 2005, the number of edges (re- peers and online-only contacts” is cited 533 times. Similarly, lationships) doubled from 2000. Moreover, Fig. 3(c) illustrates for Fig. 3(e) the connection between victimizing others and some of the serious consequences of bullying like violence, psychiatric problem is important because corresponding cita- learning disability, and depressive disorder. The research effort tion “Cross-national time trends in bullying victimization in on bullying increases significantly in the year 2010 as the cor- 33 countries among children aged 11, 13 and 15 from 2002 to responding graph contains 15 nodes and 17 edges (Fig. 3(d)). 2010” is cited 73 times. Finally, for the year 2017 (Fig. 3(f)), we Dangerous relations like completed suicide, at risk for suicide, acknowledge Victimizing others and Psychosocial problem is and retinoblastoma are occurred in this graph. Also, Fig 3(d) a vital relationship (citation title: “Consequences of bullying presents some interesting secondary relationship of bullying. victimization in childhood and adolescence: A systematic For example, bullying occurs with aggressive behavior and review and meta-analysis”, cited by 20 times). We observe aggressive behavior occurs with obesity and alcohol and other that many mental diseases and disorders are co-occurring drugs (AOD) use. In the graph for 2005, depressive disorder with bullying over the time leading to strong ties between maintains an indirect relationship with bullying (depressive bullying and mental disorders. disorder–victimizing others–violence–bullying). However, in the 2010 bullying graph depressive disorder maintains a di- 5 USER INTERFACE rect relationship with bullying. In the year 2015, the graph A Graphical User interface (GUI) can be developed on top expands by a large margin, which contains 21 nodes and of semantic MEDLINE and Google Scholar citation data for 24 edges. This graph shows a direct relation between bul- interactive data analysis. Semantic MEDLINE data is avail- lying and suicide. We consider it an important relationship able in [5]. “Scholarly” software [3] can be employed to get because numerous research studies are performed on this Google Scholar data. Next, it is possible to create a Resource relationship. A number of interesting concepts appear in the Description Framework (RDF) database based on semantic graph (Fig. 3(e)); in the following we provide some examples: MEDLINE and Google Scholar datasets and a GUI on top autism spectrum disorder, paranoia, drug abuse, narcissism, of the RDF database. RDF database allows incorporation of etc. Besides, like earlier, in the 2010 bullying graph psycholog- ontologies as well. ical distress maintains an indirect relationship with bullying We propose a web-based system called “Bullying Ana- (psychological distress–victimizing other–bullying). However, lytics” which support semantic search, indirect effect find- psychological distress maintains a direct relationship with ing, and aggregated statistics. We present system designs in bullying in 2015 graph. In 2017 graph (Fig. 3(f)), we have 21 Figs. 4, 5, 6. nodes and 21 edges. In this case, the growth is smaller as the gap between the year 2015 and 2017 is only two years. Semantic Search However, in these two recent years, interesting concepts such as family conflict, self-injurious behavior, chronic disease, Semantic Search permits users to formulate user define queries paranoid ideation, and inflammation appear in the graph. with a GUI that uses ontology-driven search. Fig. 4 shows From Semantic MEDLINE, we collect citation titles of how Semantic Search retrieves bullying literature that con- the relations available (shown in Fig. 3). We search Google tains “Female” and “Depression” related terms and the to- Scholar with these titles to find their “cited-by" count. The tal cited-by count is more than or equal 10. By using “Add cited by information is collected on April 3, 2018. The citation Statement” button, researchers can incrementally formulate with title “Bullying: its roots in pre-school aggression.” repre- a complex query which helps to find precise information. sents the single edge in the graph shown in Fig. 3(a). We did Query results (literature title, published year, and cited-by) not find this citation in Google Scholar. Fig. 3(b)’s top citation are presented in the results section. title is “Bullying behaviour and psychosocial health among school students in New South Wales, Australia: cross sec- Pathfinder tional survey”, that established the relation between bullying This service allows users to find indirect effects of bullying and psychiatric problem, has been cited 581 times. For 2005 and interactive graph-based visualization from published graph (Fig. 3(c)), we identify relationship between victim- literature. A user can define hope size as well. Fig. 5 presents izing others and depressive disorder very important because how bullying indirect effects can be retrieved where hope size the corresponding citation (title: Does bullying cause emo- is 3 and relation types are COEXISTS_WITH and CAUSES. tional problems? A prospective study of young teenagers) Users can click an edge of the graph to get detail publication is cited 1,057 times. We consider the connection between information. Moreover, users can check whether a edges in bullying and harassment are important for (Fig. 3(d)) because the path is correct or not and provide “Correct” or “Incorrect” corresponding citation “Does online harassment constitute information through “Feedback” section. Path graph updates Exploiting Literature-Based Discovery to Study Effects of Bullying IUI Workshops’19, March 20, 2019, Los Angeles, USA Figure 4: Semantic Search. Figure 6: Aggregated Statistics. based on users feedback. Pathfinder service allows users to 6 CONCLUSION AND FUTURE WORK find a long chain of indirect effect interactively. In this paper, we describe an LBD approach to study bullying. We show that the LBD approach is capable of performing advanced analytics on bullying. In this paper, we present a small number of direct and indirect consequences of bullying. Our future work include development of a software tool to find the long chain of meaningful indirect consequences automatically. It is challenging to design a software that can understand literature semantics. We also plan to use graph algorithms (e.g., PageRank) for advanced analytics. 7 ACKNOWLEDGMENTS This manuscript has been authored by UT-Battelle, LLC un- der Contract No. DE-AC05-00OR22725 with the U.S. Depart- ment of Energy. The United States Government retains and the publisher, by accepting the article for publication, ac- knowledges that the United States Government retains a non- exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Figure 5: Pathfinder. Department of Energy will provide public access to these re- sults of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe- Aggregated Statistics public-access-plan). We especially thank Dr. Maleq Khan for his useful guide- Fig. 6 shows aggregation service that “Bullying Analytics” lines and suggestions that greatly improved the paper. provides. Users can select a start and end date. Based on users selection the GUI presents the total number of bullying- REFERENCES related citations published by year as a bar chart. Moreover, [1] 2018. Facts About Bullying. URL: https://www.stopbullying.gov/ GUI presents top disease, or disorder related terms occurred media/facts/index.html. in bullying literature as a pie chart (a partical example is [2] 2018. Google Scholar. URL: https://scholar.google.com. provided in Fig. 6). [3] 2018. scholarly 0.2.4. URL: https://pypi.org/project/scholarly/. IUI Workshops’19, March 20, 2019, Los Angeles, USA Trovato and Tobin, et al. [4] 2018. Semantic MEDLINE. URL: https://skr3.nlm.nih.gov/SemMed/ [22] Judith MacIntosh. 2005. Experiences of workplace bullying in a rural jsp/SemMedDocumentation2.pdf. area. Issues in Mental Health Nursing 26, 9 (2005), 893–910. [5] 2018. Semantic MEDLINE - Database Download. URL: https://skr3. [23] Karen A Matthews, J Richard Jennings, Laisze Lee, and Dustin A Par- nlm.nih.gov/SemMedDB/download/download.html. dini. 2017. Bullying and Being Bullied in Childhood are Associated [6] Bushra Akram and Mubeen Ilyas. 2017. Coping Strategies, Mental with Different Psychosocial Risk Factors for Poor Physical Health in Health and HIV Status: Predictors of Suicidal Behaviour among PWIDs. Men. Psychological Science 28, 6 (2017), 808–821. JPMA 67, 568 (2017). [24] Christopher M Miller, Thomas C Rindflesch, Marcelo Fiszman, Dimitar [7] Mikael Andersén and Andreas Birgegård. 2017. Diagnosis-Specific Hristovski, Dongwook Shin, Graciela Rosemblat, Han Zhang, and King- Self-Image Predicts Longitudinal Suicidal Ideation in Adult Eating man P Strohl. 2012. A Closed Literature-Based Discovery Technique Disorders. International Journal of Eating Disorders 50, 8 (2017), 970– Finds a Mechanistic Link Between Hypogonadism and Diminished 978. Sleep Quality in Aging Men. Sleep 35, 2 (2012), 279–285. [8] Onesemus Awiria. 1994. Bullying at School-What We Know and What [25] Tomoyo Mitsui, Toshiyuki Yoshida, and Gen Komaki. 2017. Psycho- We Can Do. metric Properties of the Eating Disorder Examination-Questionnaire [9] Esther Calvete, Liria Fernández-González, Joaquín M González- in Japanese Adolescents. BioPsychoSocial Medicine 11, 1 (2017), 9. Cabrera, and Manuel Gámez-Guadix. 2018. Continued Bullying Victim- [26] Dan Olweus. 1973. Hackkycklingar och översittare: Forskning om ization in Adolescents: Maladaptive Schemas as a Mediational Mecha- skolmobbning. nism. Journal of Youth and Adolescence 47, 3 (2018), 650–660. [27] Dan Olweus. 1994. Bullying at School. In Aggressive Behavior. Springer, [10] CDC. 2014. The Relationship between Bullying and Suicide: What 97–130. We Know and What it Means for Schools. Available on the In- [28] Naomi Priest, Tania King, Laia Bécares, and Anne M Kavanagh. 2016. ternet: http://www. cdc. gov/violenceprevention/pdf/bullying-suicide- Bullying Victimization and Racial Discrimination among Australian translation-final-a. pdf (2014). Children. American Journal of Public Health 106, 10 (2016), 1882–1884. [11] Chiara Conti, Chiara Mennitto, Giulia Di Francesco, Federica Fraticelli, [29] Charlotte Rayner. 1997. The Incidence of Workplace Bullying. Journal Ester Vitacolonna, and Mario Fulcheri. 2017. Clinical Characteristics of Community & Applied Social Psychology 7, 3 (1997), 199–208. of Diabetes Mellitus and Suicide Risk. Frontiers in Psychiatry 8 (2017), [30] Thomas C Rindflesch, Halil Kilicoglu, Marcelo Fiszman, Graciela 40. Rosemblat, and Dongwook Shin. 2011. Semantic MEDLINE: An Ad- [12] D. R. Swanson. 1988. Migraine and Magnesium: Eleven Neglected vanced Information Management Application for Biomedicine. Infor- Connections. Perspectives in Biology and Medicine 31, 4 (1988), 526– mation Services & Use 31, 1-2 (2011), 15–21. 557. [31] Tanveer Singh, Neetu Bagga, Anureet Kaur, Navjot Kaur, Dinesh Yu- [13] Michael D Gordon and Robert K Lindsay. 1996. Toward Discovery graj Gawande, and Rajesh Kumar Goel. 2017. Agmatine for Com- Support Systems: A Replication, Re-Examination, and Extension of bined Treatment of Epilepsy, Depression and Cognitive Impairment in Swanson’s Work on Literature-Based Discovery of a Connection Be- Chronic Epileptic Animals. Biomedicine & Pharmacotherapy 92 (2017), tween Raynaud’s and Fish Oil. JAIST 47, 2 (1996), 116–128. 720–725. [14] Amy L Gower, Molly Cousin, and Iris W Borowsky. 2017. A Multilevel, [32] Jeremy G Stewart, Linda Valeri, Erika C Esposito, and Randy P Auer- Statewide Investigation of School District Anti-Bullying Policy Quality bach. 2018. Peer Victimization and Suicidal Thoughts and Behaviors and Student Bullying Involvement. Journal of School Health 87, 3 (2017), in Depressed Adolescents. Journal of Abnormal Child Psychology 46, 3 174–181. (2018), 581–596. [15] Melissa K Holt, Jennifer Greif Green, Mina Tsay-Vogel, Joanna David- [33] Don R Swanson. 1986. Fish Oil, Raynaud’s Syndrome, and Undiscov- son, and Claire Brown. 2017. Multidisciplinary Approaches to Research ered Public Knowledge. Perspectives in biology and medicine 30, 1 on Bullying in Adolescence. Adolescent research review 2, 1 (2017), (1986), 7–18. 1–10. [34] Swanson, Don R. 1978. Aggression in the Schools: Bullies and Whipping [16] Jun Sung Hong, Dexter R Voisin, Sujung Cho, Douglas C Smith, and Boys. Hemisphere. Stella M Resko. 2017. Peer Victimization and Substance Use Among [35] Rebekah L Thomas, Adrian B Kelly, Gary CK Chan, and Joanne W African American Adolescents and Emerging Adults on ChicagoâĂŹs Williams. 2017. Linking Eating and Weight Control Attitudes to Re- Southside. (2017). lationship Experiences: A Large-Scale Survey of Adolescent Females. [17] Dimitar Hristovski, Carol Friedman, Thomas C Rindflesch, and Borut Journal of Adolescence 58 (2017), 67–73. Peterlin. 2006. Exploiting Semantic Relations for Literature-Based [36] Anthony A Volk, René Veenstra, and Dorothy L Espelage. 2017. So You Discovery. In AMIA symposium, Vol. 2006. AMIA, 349. Want to Study Bullying? Recommendations to Enhance the Validity, [18] Jaana Juvonen and Sandra Graham. 2014. Bullying in Schools: The Transparency, and Compatibility of Bullying Research. Aggression and Power of Bullies and the Plight of Victims. Annual Review of Psychology Violent Behavior 36 (2017), 34–43. 65 (2014), 159–185. [37] US WBI. 2014. Workplace Bullying Survey. Workplace Bullying Institute [19] Halil Kilicoglu, Marcelo Fiszman, Alejandro Rodriguez, Dongwook 8, 2014 (2014), 6. Shin, A Ripple, and Thomas C Rindflesch. 2008. Semantic MEDLINE: A [38] Han Zhang, Marcelo Fiszman, Dongwook Shin, Bartlomiej Wilkowski, Web Application for Managing the Results of PubMed Searches. 2008 and Thomas C Rindflesch. 2013. Clustering Cliques for Graph-Based (2008), 69–76. Summarization of the Biomedical Research Literature. BMC Bioinfor- [20] Iliana Kotwas, Aileen McGonigal, Mireille Bastien-Toniazzo, Fabrice matics 14, 1 (2013), 182. Bartolomei, and Jean-Arthur Micoulaud-Franchi. 2017. Stress Reg- [39] Xiaoyan Zhang, Chaelin Karen Ra, Donglan Zhang, Yunting Zhang, ulation in Drug-Resistant Epilepsy. Epilepsy & Behavior 71 (2017), and Kara E MacLeod. 2016. The Impact of School Social Support and 39–50. Bullying Victimization on Psychological Distress among California [21] Nicole Krämer, Sabrina Sobieraj, Dan Feng, Elisabeth Trubina, and Adolescents. Californian Journal of Health Promotion 14, 2 (2016), 56. Stacy Marsella. 2018. Being Bullied in Virtual Environments: Experi- ences and Reactions of Male and Female Students to a Male or Female Oppressor. Frontiers in Psychology 9 (2018), 253.