Page 140 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 42
P. 140

OU Shiyan, TANG Zhengui & SU Feifei / Construction and usage of terminology services for information retrieval  139


                                                             Terminology
                     Goal A                                 Service Usage
                                                              Mode(A)


                                    Effectiveness              Efficiency  User satisfaction  Learnability
                    Criteria B
                                       (B1)                      (B2)      (B3)       (B4)
                              Representation  Improvement  Freedom of  Operation  Clicking
                  Sub-Criteria Bi degree of retrieval  degree of retr-  term selec-
                             requirement(B11) ieval results(B12)  tion(B13)  time(B21) times(B22)



                              Mode 1˖Multiple  Mode 2˖Single   Mode 3˖Hybrid of  Mode 4˖Automa-
                  Alternatives C  choice for query  choice for query   query expansion and  tic query expansion
                                expansion (C1)  replacement  (C2)  replacement(C3)  (C4)
                            Figure 13. AHP hierarchy for terminology service usage mode selecting decision.

               results have no much difference (Gao, 2007). Here we used the latent root method to calculate
               the weight of each evaluation criterion (shown from Table 4 to Table 6). During calculating the
               weights, a consistency ratio, which expresses the internal consistency of the judgments, was also
               checked. In the above three decision matrixes, all the consistency ratios are less than 0.1 and thus
               acceptable. According to the calculated weights, we obtained the usability evaluation index system
               as shown in Table 7. Among these seven criteria, except for operation time and clicking times, the
               other five criteria were judged subjectively based on 5-point Likert Scale. While calculating the
               total usability score, to keep the measures of all the criteria consistent, we discretized the observed
               values of operation time and clicking times to convert them into 5-point scales.


               Table 3. The fundamental importance for pairwise comparisons
                 Intensity of   1          3          5          7           9       2、4、6、8
                 importance
                              Equal     Moderate    Strong    Very strong  Extreme   Intermediate
                  Definition
                             importance  importance  importance  importance  importance  values
               Note:If the importance ratio between the element i and j is aij, then the importance ratio between the element j and i is
               aji=1/aij

               Table 4.  Judgements in the decision matrix of the top criteria
                    A           B1          B2          B3          B4           Weight (%)

                    B1          1           2           1/2         2              26.3
                    B2          1/2         1           1/3         1              14.1
                    B3          2           3           1           3              45.5

                    B4          1/2         1           1/3         1              14.1
               Note:Consistency Ratio (CR) =0.4 %< 0.1
   135   136   137   138   139   140   141   142   143   144   145