홈으로ArticlesAll Issue
ArticlesCoNavigator: A Framework of FCA-Based Novel Coronavirus COVID-19 Domain Knowledge Navigation
  • Fei Hao 1,2 and Doo-Soon Park3,*

Human-centric Computing and Information Sciences volume 11, Article number: 06 (2021)
Cite this article 5 Accesses
https://doi.org/10.22967/HCIS.2021.11.006

Abstract

The rapidly increasing number of confirmed cases and extremely large amounts ofmedical resources about coronavirus disease 2019 (COVID-19)pandemic are enriching theCOVID-19 domain knowledge bases. Undoubtedly, it is resulting in “informationoverload” and “information confusion” for users. Hence, how to achieve precision information navigation for COVID-19 related domain knowledge retrievaland personalized services, has become a huge challenge of this research. Due tothe unique retrieval and navigation advantages of concept lattice, a framework ofnovel coronavirus COVID-19 domain knowledge navigation named CoNavigator,based on Formal Concept Analysis (FCA) theory is devised. It provides apromising approach for improving the retrieval efficiency from various novelcoronavirus COVID-19 information resource platforms. Besides, some usefulpatterns are extracted and regarded as important evidence for an appropriate understanding of the pathological features of novel coronavirus COVID-19.


Keywords

COVID-19, FCA, Concepts, Navigation


Introduction

Coronavirus disease 2019 (COVID-19) pandemic is caused by an emerging newcoronavirus (severe acute respiratory syndrome coronavirus-2 [SARS-CoV-2]). It mainly exhibits some typical symptoms, such as respiratorydistress, cough, and low fever. From December 2019 to July 2020, the COVID-19 pandemic has infected a reported 13,575,158 number of people, resulting in 584,940 deaths, in 216 countries and areas around the world, andcontinue to spread exponentially [1]. In most countries, this pandemic cannot becompletely controlled and is still going on.
At present, many professional scholars and scientific research institutions haveconducted research and analysis on the new coronavirus analysis. China has basicallymastered the control strategy and measures of the COVID-19 inetiology, epidemiological characteristics, and pathogenic mechanism [24], etc., andprovided an important reference for the global community to understand the essentialsof CVOID-19. With the rapid development of big data technologies, massive amounts of COVID-19 related data are constantly being generated, whichleads to difficulty inobtaining accurate information resources. Currently,there exist some COVID-19 domain knowledge search and query platforms[5, 6],but these searches are conducted for obtaining the COVID-19 literature based ontraditional retrieval models, which are inefficient and difficult to satisfy the long-termcritical information resource retrieval requirements. To address this challenge,the FCA technique could be an alternative for building an efficient information retrievalsystem [7, 8]. Codocedoand Napoli [9] revealed that FCA-based information retrieval isbetter than the traditional way since it has the advantage of searching, and retrievesmore satisfactory results from users. Qadi et al.[10] proposed an information retrievalsystem based on FCA. Specifically, their system allows users to navigate the hierarchicalformal concepts for retrieving the relevant documents regarding the user’squery. FCA, a powerful data analysis methodology, has been broadly adopted invarious fields, such as cognitive learning[11], social network analysis[[12], [13]], diseasediagnosis[14], search engine[15], and so forth.
With the above advantages and extensibility of FCA, especially the formal concepts could better characterize the potential relations between objects and attributes (the relations in this work refer to the binary relations between patients and symptoms), this research attempts to study the COVID-19 domain knowledge navigation and then further provides the relevant intelligent services for medical doctors andpatients. The specific contributions of this work are outlined as follows.
We present a framework of FCA-based novel coronavirus COVID-19 domainknowledge navigation, called CoNavigator. Specifically, we construct the formalcontext of COVID-19 patient medical recordsby taking the patients as the objects and the symptoms as the attributes and then generate the concept latticeaccordingly. We store the obtained concepts in a database for supportingthe further medical query request.
Considering the dynamicity of the formal context of COVID-19 patient medicalrecords, we present an object-incremental concept lattice generation algorithmfor obtaining the concepts more efficiently. Our proposed incrementalalgorithm is to preserve the previously obtained concept lattice and generatethe concept lattice of the added formal context. Then, we make the intersectionof them and store it into a set. Finally, the extent for each intent inthis set can be obtained. The experiment shows that our algorithm canobtain the concepts faster than other existing non-incremental algorithms.
We conduct a case study for evaluating the query efficiency and effectivenessofthe proposedFCA-basednovel coronavirus COVID-19 domain knowledge navigation model. Importantly,some useful patterns are extracted and analyzed for explaining the pathologicalcharacteristics of novel coronavirus COVID-19 and providingpowerful evidence for further investigation on this disease.
The outline of this paper is as follows. The preliminary knowledge of FCA isprovided in Section 2. Then, Section 3 provides a framework of FCA-based novelcoronavirus COVID-19 domain knowledge navigation. Section 4 conducts a concretecase study for evaluating the efficiency of our concept lattice generation algorithmas well as extracting more useful patterns for better explanations on pathologicalfeatures of novel coronavirus COVID-19. Finally, Section 5 concludes this work.


The Basics of FCA

FCA, invented by Rudolf Wille in 1982[16], is often used to characterize the relationships between concepts in a certain domain and understand the data by representingit as a concept lattice. This section will briefly overview the basics of FCAmethodology[17].
Four core definitions in FCA including formal context, formal concept,partial order, and conceptlatticeare given as follows.
1. A formal context is formed as a 3-tupleFC=(U,M,I), where U andMrefer to a set of objects and attributes respectively, andIis the binary relation between U and M (i.e.,I⊆U⨂M). Suppose u∈U and m∈M, each (u,m)∈Iis interpreted as the object u has the attribute m, otherwise (u,m)∉I.

2. A pair (X,Y) is a formal concept of the formal context FC=(U,M,I),if pair (A,B)(A⊆U,B⊆M)meets A^↑=Band B^↓=A, then, Aand Bare called the extent and intent of the concept (A,B). Note that, the operators ↑ and↓on A⊆U,B⊆Mfor =(U,M,I), are defined as follows.

(1)

(2)


3. SupposeC(FC) be the set of all formal concepts obtained from FC=(U,M,I).If (A1,B1 ),(A2,B2)∈C(FC), then let

(3)

then “≤” denotes a partial relation of C(FC).

4. A concept lattice can be constructed by organizing all formal concepts C(FC)of a context, FC with the partial order≤denoted as L=(C(FC),≤). Giventwo concepts C1=(A1,B1) and C2=(A2,B2), ifA1⊂A2 and B2⊂B1 then a partialorderC1⊂C2 holds and there does not exist C3 so thatC1≤C3≤C2, then thereexists an edge between concepts C1 and C2. Usually, we call C1 is the son-conceptof C2, and C2 is the father-concept of C1.

Example 1.Fig. 1 shows an example ofa document-keywords binary relationaltable and its corresponding concept lattice. As we can see, the set of documents,i.e.,U={d1,d2,…d5}, and the set of keywords, i.e., M={a,b,c,d,e}.
U/M a b c d e
d1 × ×
d2 × × ×
d3 × ×
d4 × ×
d5 × ×
(a)
(b)
Fig. 1.(a) Formal context and (b) concept lattice.

Clearly, we finally obtain 7 concepts. For example, the concept ({d1,d2 },{a,b})implies that documents d1 and d2 have the common keywords aandb. Therefore, theconcept lattice can be viewed as a kind of clustering results for documents in terms of keywords.


FCA-Based COVID-19 Domain Knowledge Navigator

In this section, we present an FCA-based novel-coronavirus COVID-19 domainknowledge navigator, CoNavigator.First of all, the architecture of CoNavigator isprovided. Then, we will elaborate each module of this architecture including formal context construction, concept lattice generation as well as concept lattice storageand retrieval.

The Architecture of CoNavigator
Fig. 2 showsthe architecture of CoNavigator. It is composed of three functionalitymodules: (1) formal context construction; (2) concept lattice generation; and (3) conceptlattice storage and retrieval.
As can be seen from Fig. 2, the COVID-19 patient medical records are the input of CoNavigator, then we first construct the formal context of COVID-19 patient medical records; afterward, a corresponding concept lattice is generated; the obtained formal concepts included in the concept lattice are stored in a database for serving the users’ query request; with the query request, the query results will be retrieved and returned to the users.
Fig. 2. The Architecture of CoNavigator.

Formal Context Construction
Considering the patient medical records for novel coronavirus COVID-19 [18] as shown in Table 1, we construct the formal context FC=(PA,S,I), by taking the patients as the objects P, and the inherent attributes (i.e., age, sex, and location) as well as the medical attributes (i.e., symptoms, travel history and chronic disease) as the attributes S. And,Iindicates a relationship between the patient and its attributes.

Table 1. Patient medical records for novel coronavirus COVID-19
Since the patient medical dataset contains several multi-value attributes, such as the symptoms of a given patient is composed of {“cough”,“fatigue”,“fever”,“sputum”, “myalgias”,“shortness_of_breath”,“respiratory_symptoms”,“diarrhea”,“rhinorrhea”,“chest_pain”,“sore_throat”,“gasp”,“pneumonitis”, “dizziness”, “fatigue”, “headache”}.We utilized vector encoding to represent multi-value attributes. For example, if patient A has symptoms of “shortness_of_breath” and “chest_pain”, then A’s symptomscan be represented with a vector as follows,

(4)

Therefore, the formal context of the above patient medical dataset is shown inTable 2.

Table 2. The formal context of patient medical dataset
Age Range 1 Age Range 2 Age Range 3 Age Range 4 Sex male Sexfemale Travel Loc. 1 Travel Loc. 2 Travel Loc. k Symptom 1 Symptom 2 Symptom m
PA1 × × × × ×
PA2 × × × × × ×
× × × × ×
PAn × × × × ×

Concept Lattice Generation
This section is devoted to incremental generation on the concept lattice of COVID-19 formal context. Since the number of novel coronavirus COVID-19 patients is rapidly increasing,the above constructed formal context FC is dynamically updated. It will lead to a great challenge for quickly generating the concept lattice. In this paper, the objectincrementalconcept lattice generation algorithm is devised for better addressing this challenge and further obtaining the corresponding concept lattice L(FC).
Let us consider an initial formal context FC1=(PA1,S,I1) (here, S={a,b,c,d,e,f,g,h,i,j,k,l} can be dynamically updated by adding kpatients and their symptoms FC2=(PA2,S,I2) (pink area), as shown in Table 3, and finally reach the resulting formal context FC=(PA,S,I), where PA=PA1∪PA2,I=I1∪I2. The solution idea is that: our algorithm is to preserve the previously obtained concept lattice L(FC1)and generate the concept lattice L(FC2) of the added formal context FC2. Further, the intersection of C(FC1) and C(FC2) is operated, i.e., (FC1)∩C(FC2),and store it into a setC(FC). After that, we obtain the extent efor each intenti∈C(FC)via e←i^ sup>↓.

Table 3. Object-incremental formal contexts for adding COVID-19 patients
PA × S a b c d e f g h i j k l
PA1 × × ×
PA2 × × ×
× × × × ×
PAn × × × ×
PAn+1 × × × × ×
PAn+2 × × × × ×
× × × × × ×
PAn+k × × × ×
Theorem 1. Given three given formal contexts FC1=(PA1,S,I1), FC2=(PA2,S,I2), and FC=(PA1∪PA2,S,I1∪I2),the relation among the set of the intents C(FC1),C(FC2), and C(FC)satisfies the following equation:

(5)

Proof. This proof is similar to the attribute-oriented concept lattice generationalgorithm[19]. We will not elaborate on it due to the paper length limitation.

Concept Lattice Storage and Retrieval
In order to achieve information navigation and retrieval, we store the generated concept lattice into a relational database where the fields are extent, intent,andparent. Note that the parent refers to the father concept of the current concept. In thisdatabase, each record indicates a concept.
We adopt the following SQL scripts to create the above relational database andtable concepts.

CREATEDATABASECOVID;
CREATETABLEConcepts(
ConceptID int,
Extent varchar(255),
Intent varchar(255),
Parent varchar(255),
);

After storing the concept lattice with the relational database, we can achieve accurate retrieval of useful patterns, such as common symptoms, of COVID-19patients. Based on the created table concepts, the implementation sentence of alluseful patterns that satisfy the user’s search keywords.

SELECTExtent FROMConcepts WHEREIntent= 'keywords'

Note that “keywords”are typed by users, such as “cough” and “chest_pain”.


Case Study

In this section, for better illustrating our proposed approach, we consider 20 COVID-19 patients and their 14 symptoms only, then the useful patterns will bedistilled from the concept lattice.

Setup
Regarding the above dataset, we take 20 patients as the objects, i.e., PA={PA1,PA2,… PA20}, and 14 symptoms as the attributes, i.e., S={cough,fatigue,fever,sputum,myalgias,shortness_of_breath,respiratory symptoms,chest_pain,diarrhea,rhinorrhea,sneezing,sore throat,gasp,weakness}.For better representation in the constructed formal context, we useS_i (i=1,2..,14) to denote thedifferent symptoms (as depicted in Table 4).

Table 4. Formal context of case study data
PA × S S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14
PA1 × × × × ×
PA2 × ×
PA3 × × × ×
PA4 × × ×
PA5 ×
PA6 × × ×
PA7 × × ×
PA8 × × ×
PA9 × × ×
PA10 × ×
PA11 × × × × ×
PA12 × × × × ×
PA13 × × × × ×
PA14 × × × × ×
PA15 ×
PA16 × × ×
PA17 × × ×
PA18 × × × ×
PA19 × × ×
PA20 × × ×
According to the concept lattice generation algorithm, the following concept lattice is obtained as shown in Fig.3. Each node is corresponding a concept in the conceptlattice. For example, a concept ({PA7, PA8},{s3,s6,s8})can be interpreted as patients PA7, PA8 have the common symptoms of fever, shortness of breath, andchest pain.
Fig. 3. Concept lattice of formal context of case study data.


Table 5. Concepts relational table Concepts
Concept ID Extent Intent Parent
1 {PA1, PA3, PA4, PA6, PA7, PA8, PA9, PA10, PA11, PA12, PA13, PA14, PA16, PA17, PA18, PA19, PA20} {S3} inf
2 {PA1, PA3, PA15} {S2} inf
3 {PA2, PA7, PA8, PA9} {S6} inf
4 {PA1, PA3, PA4, PA5, PA6, PA9, PA10, PA11, PA12, PA16, PA17, PA18, PA19, PA20} {S1} inf
5 {PA11, PA12, PA13, PA14} {S3, S9, S10} #1
6 {PA13, PA14, PA16, PA17} {S3, S12} #1
7 {PA3, PA6, PA7, PA8, PA13, PA14, PA18} {S3, S8} #1
8 {PA7, PA8, PA9} {S3, S6} #1, #3
9 {PA1, PA3, PA4, PA6, PA9, PA10, PA11, PA12, PA16, PA17, PA18, PA19, PA20} {S1, S3} #1, #4
10 {PA2} {S6, S7} #3
11 {PA13, PA14} {S3, S8, S9, S10, S12} #5, #6, #7
12 {PA7, PA8} {S3, S6, S8} #7, #8
13 {PA11, PA12} {S1, S3, S9, S10, S11} #5, #9
14 {PA16, PA17} {S1, S3, S12} #6, #9
15 {PA3, PA6, PA18} {S1, S3, S8} #7, #9
16 {PA9} {S1, S3, S6} #8, #9
17 {PA19, PA20} {S1, S3, S14} #9
18 {PA1, PA3} {S1, S2, S3} #2, #9
19 {PA1, PA4} {S1, S3, S4} #9
20 {PA18} {S1, S4, S8, S13} #15
21 {PA3} {S1, S2, S3, S8} #15, #18
22 {PA1} {S1, S2, S3, S4, S5} #18, #19


For the sake of storage, we insert all obtained concepts into the table Concepts of COVIDdatabase.
Table 5 is composed of all obtained formal concepts including concept ID, extent,intent, and parent concept. For instance, for No.10 formal concept({PA2},{s6, s7}), its parent concept is No.3 concept. Based on Table 5, we canextract some useful patterns by a SQL query. For example, if we want to search the CVOID-19 patients who have the symptom of shortness of breath, the SQL script isas follows.

SELECTExtent FROMConcepts WHEREIntent='shortness of breath';

Consequently, the query results {PA2, PA7, PA8, PA9} will be returned to the currentuser.

Efficiency and Effectiveness Evaluation
As we know, the COVID-19 patient medical database is not static and can be updatedsince the confirmed new patients will be added to the database every day.As we mentioned before, how to update our concepts relational table incrementallyand respondto the query results quickly is interesting and challenging research.
Let us assume that 15 new confirmed COVID-19 patients are added to our medicaldatabase, we also conduct an efficiency evaluation on obtaining the conceptsby comparing our incremental algorithm with other two existing algorithms:CMCG algorithm[20] and non-incremental algorithm in terms of running time. Fig.4 shows the comparison of running time for generating concept lattice.That is to say, our proposed incremental algorithm can significantly save runningtime. Therefore, it is suitable and efficient for handling the dynamical novel coronavirusCOVID-19 patient medical records.
Fig. 4. Comparison of running time for generating concept lattice.


If we do not rely on the concept lattice structure to achieve useful patterns retrieval,the relational table including fields,patient and symptoms, is shown in Table6.
Regarding Table 6, if we want to query the patients who satisfy the inputkeywords, the SQL script is given as follows.

SELECTExtent FROMPatientKey WHEREKeys LIKES'%Keywords%';

Obviously, in a large COVID-19 patient medical database, the effectiveness ofan accurate query is higher than the fuzzy query.

Table 6. Non-concepts relational table PatientKey
ID Patient Symptom
1 PA1 S1, S2, S3, S4, S5
2 PA2 S6, S7
3 PA3 S1, S2, S3, S8
4 PA4 S1, S3, S4
5 PA5 S1
6 PA6 S1, S3, S8
7 PA7 S3, S6, S8
8 PA8 S3, S6
9 PA9 S1, S3, S6
10 PA10 S1, S3
11 PA11 S1, S3, S9, S10, S11
12 PA12 S3, S8, S9, S10, S11
13 PA13 S3, S8, S9, S10, S12
14 PA14 S3, S8, S9, S10, S12
15 PA15 S2
16 PA16 S1, S3, S12
17 PA17 S1, S3, S12
18 PA18 S1, S3, S8, S13
19 PA19 S1, S3, S14
20 PA10 S1, S3, S14

Useful Patterns Extraction and Analysis
According to the obtained concepts, we can summarize the following useful patternsby observing the hidden relationships between extent and intent.
(Top-k 1-intent concepts). Intuitively,the top-k 1-intent concept is usedto characterize the importance of attributes[21].
For example, top-2 1-intent concepts ({PA1, PA3, PA4, PA6… PA14, PA16,…, PA20},{s3})and ({PA1,…, PA6, PA9, PA10, PA11, PA12, PA16,…, PA20},{s1} )reflectthe attributes s1 and s3 are very important since they are owned by manypatients. In our case study, it is easy to find that more than 65% of patientshave symptoms of cough and fever. It coincides with the official reportson major symptoms of novel coronavirus COVID-19, released fromthe World Health Organization (https://www.who.int/health-topics/coronavirus#tab=tab_1).
(Frequent attributes in concepts). By observing the intents in all concepts,we find that attribute s3 is the most frequently appeared symptom among thepatients. That is to say, the fever symptom is a critical signal for identifying apatient if he/she suffers from the novel coronavirus COVID-19. That is why manyfever clinics are deployed during the pandemic of COVID-19.
(Non-trivial attributes in concepts). Some special patients with chronicdiseases usually have some non-trivial symptoms compared to normalpeople. Technically, this pattern can be extracted by mining the non-trivialtributes in concepts. For example, the attribute s7is a non-trivial attributesince it only appears in No.10 Concept. And the extent of No.10 Concept is PA2, it implies that patient PA2 as old people are more susceptible to new coronaviruspneumonia with symptom respiratory symptoms compared to youngpatients.
By analyzing the obtained patterns, it is useful to help medical doctors to masterthe pathological features of novel coronavirus COVID-19. Besides, thesepatterns can be used for constructing the knowledge graph, domain search engine.


Conclusion

At present, the massive patient data generated by the ongoing novel coronavirus CVOID-19 is leading to the information overload problem to scientists who are working in the field of COVID-19. Thus, it is becoming difficult to extract more useful patterns from novel coronavirus CVOID-19 patient medical records. Towards this end, this research develops CoNavigator, a framework of FCA-based novelcoronavirus CVOID-19 domain knowledge navigation. With this framework, some useful patterns are extracted for supporting the accurate and intelligent query andfurther providing more appropriate decisions for users. However, the proposed approach has a limitation for deploying the algorithm on a large-scale coronavirusCVOID-19 patient medical records due to the expensive concept generation time. Infuture work, we will further optimize the algorithm for concept generation inorder to achieve better information navigation.


Acknowledgements

None.


Author’s Contributions

Methodology,FH, DSP.Writing review and editing, FH, DSP. Resources, DSP. Software, DSP. Supervision, DSP. Writingoriginal draft, FH.


Funding

This work was funded in part by the National Natural Science Foundation of China (No. 61702317), the NationalResearch Foundation of Korea (No. NRF-2020R1A2B5B01002134), BK21 FOUR(Fostering Outstanding Universities for Research)(No.5199990914048), and the Fundamental Research Funds for the Central Universities, China (No. GK202103080).


Competing Interests

The authors declare that they have no competing interests.


References

[1] G. Livadiotis, “Statistical analysis of the impact of environmental temperature on the exponential growth rate of cases infected by COVID-19,” PLoS One, vol. 15, no. 5, article no. e0233875, 2020. https://doi.org/10.1371/journal.pone.0233875
[2] S. Chen, J. Yang, W. Yang, C. Wang, and T. Barnighausen, “COVID-19 control in China during mass population movements at New Year,” The Lancet, vol. 395, no. 10226, pp. 764-766, 2020.
[3] Z. Y. Zu, M. D. Jiang, P. P. Xu, W. Chen, Q. Q. Ni, G. M. Lu, and L. J. Zhang, “Coronavirus disease 2019 (COVID-19): a perspective from China,” Radiology, vol. 296, no. 2, pp. E15-E25, 2020. https://doi.org/10.1148/radiol.2020200490
[4] Q. Ruan, K. Yang, W. Wang, L. Jiang, and J. Song, “Clinical predictors of mortality due to COVID-19 based on an analysis of data of 150 patients from Wuhan, China,” Intensive Care Medicine, vol. 46, no. 5, pp. 846-848, 2020.
[5] A. Esteva, A. Kale, R. Paulus, K. Hashimoto, W. Yin, D. Radev, and R. Socher, “Co-search: Covid-19 information retrieval with semantic search, question answering, and abstractive summarization,” 2020 [Online]. Available: https://arxiv.org/abs/2006.09595.
[6] E. Zhang, N. Gupta, R. Tang, X. Han, R. Pradeep, K. Lu, et al. “Covidex: Neural ranking models and keyword search infrastructure for the covid-19 open research dataset,” 2020 [Online]. Available: https://arxiv.org/abs/2007.07846.
[7] S. Albahli and A. Melton, “TripleFCA: FCA-based approach to enhance semantic web data management,” in Proceedings of 2016 IEEE 40th Annual Computer Software and Applications Conference (COMPSAC), Atlanta, GA, 2016, pp. 625-630.
[8] N. Juniarta, M. Couceiro, A. Napoli, and C. Raissi, “Sequential pattern mining using FCA and pattern structures for analyzing visitor trajectories in a museum,” in Proceedings of the 14th International Conference on Concept Lattices and Their Applications (CLA), Olomouc, Czech Republic, 2018.
[9] V. Codocedo and A. Napoli, “Formal concept analysis and information retrieval: a survey,” in Formal Concept Analysis. Cham, Switzerland: Springer, 2015, pp. 61-77.
[10] A. E. Qadi, D. Aboutajedine, and Y. Ennouary, “Formal concept analysis for information retrieval,” 2010 [Online]. Available: https://arxiv.org/abs/1003.1494.
[11] Y. Zhao, J. Li, W. Liu, and W. Xu, “Cognitive concept learning from incomplete information,” International Journal of Machine Learning and Cybernetics, vol. 8, no. 1, pp. 159-170, 2017.
[12] F. Hao, D. S. Park, and Z. Pei, “When social computing meets soft computing: opportunities and insights,” Human-centric Computing and Information Sciences, vol. 8, article no. 8, 2018. https://doi.org/10.1186/s13673-018-0131-z
[13] F. Hao, Z. Pei, and L. T. Yang, “Diversified top-k maximal clique detection in Social Internet of Things,” Future Generation Computer Systems, vol. 107, pp. 408-417, 2020.
[14] Y. Yuan, W. Chen, L. Yan, B. Huang, and J. Li, “A similarity-based disease diagnosis system for medical big data,” Journal of Medical Imaging and Health Informatics, vol. 7, no. 2, pp. 364-370, 2017.
[15] E. Negm, S. AbdelRahman, and R. Bahgat, “PREFCA: a portal retrieval engine based on formal concept analysis,” Information Processing & Management, vol. 53, no. 1, pp. 203-222, 2017.
[16] R. Wille, “Restructuring lattice theory: an approach based on hierarchies of concepts,” in Ordered Sets. Heidelberg, Germany: Springer, 1982, pp. 314-339. https://doi.org/10.1007/978-94-009-7798-3_15
[17] F. Hao, G. Min, Z. Pei, D. S. Park, and L. T. Yang, “K-clique community detection in social networks based on formal concept analysis,” IEEE Systems Journal, vol. 11, no. 1, pp. 250-259, 2017.
[18] Wolfram Research, “Patient medical data for novel coronavirus COVID-19,” 2020 [Online]. Available: https://doi.org/10.24097/wolfram.11224.data.
[19] Y. Yang, F. Hao, B. Pang, K. Qin, Z. Pei, and B. Li, “A quick algorithm on generating concept lattice for attribute-incremental streaming data,” in Proceedings of 2019 IEEE 21st International Conference on High Performance Computing and Communications; IEEE 17th International Conference on Smart City; IEEE 5th International Conference on Data Science and Systems (HPCC/SmartCity/DSS), Zhangjiajie, China, 2019, pp. 2811-2816.
[20] Z. Chen, N. Zhang, and Z. Wang, “New algorithm of generating concept lattice based on concept-matrix,” Computer Science, vol. 37, no. 9, pp. 180-183, 2010.
[21] F. Hao, X. Xinchang, and D. S. Park, “FCA-based θ-iceberg core decomposition in graphs,” Journal of Ambient Intelligence and Humanized Computing, 2017. https://doi.org/10.1007/s12652-017-0649-3

About this article
Cite this article

Fei Hao 1,2 and Doo-Soon Park3,*, CoNavigator: A Framework of FCA-Based Novel Coronavirus COVID-19 Domain Knowledge Navigation, Article number: 11:06 (2021) Cite this article 5 Accesses

Download citation
  • Recived8 August 2020
  • Accepted5 January 2021
  • Published15 February 2021
Share this article

Anyone you share the following link with will be able to read this content:

Provided by the Springer Nature SharedIt content-sharing initiative

Keywords