Abstract: 【Purpose/significance】By analyzing the cross-community and cross-disciplinary differences in knowledge exchange effi?
ciency in Q&A interactions, we provide references for different types of virtual communities to accurately position and optimize their
knowledge exchange support systems【. Method/process】We selected virtual communities such as ResearchGate, Zhihu, muchong.com, CSDN, and bbs.pinggu.org and their two disciplinary sections of programming languages and finance, combined with the comprehen? sive measure of community disciplinary Q&A calculated by the entropy weight method, established an index system for evaluating the efficiency of knowledge exchange through Q&A interaction in multivariate virtual communities, calculated the relative knowledge ex? change efficiency of community disciplines by using the data envelopment analysis method, and conducted cross-community and cross-disciplinary comparisons.【Result/conclusion】The overall scale efficiency of diversified virtual communities is higher, but the pure technical efficiency needs to be improved, and the marginal effect of community scale on the knowledge exchange efficiency of vir? tual communities is diminishing; at the cross-community level, the knowledge exchange efficiency of comprehensive communities is higher than that of vertical communities; at the cross-discipline level, the knowledge exchange efficiency of economic and management disciplines is higher than that of scientific and technical disciplines.【Innovation/limitation】This study breaks through the perspective of disciplinary differences, starts from the essential differences of community types, and carries out the analysis of the differences of Q&A communication efficiency between different types of communities, in order to guide the accurate positioning and optimize the con? struction mode of different types of virtual communities. Based on this study, the discipline can be further expanded and semantic and sentiment analysis can be performed on this basis.