【目的/意義】探討學術期刊影響力指數與影響因子等傳統期刊計量指標的相關性并構建該指標值的預測模型。【方法/過程】首先以圖情領域19種核心期刊為研究對象,以SPSS16.0為分析工具對影響力指數與22種傳統期刊計量指標的相關性進行分析,得到與之顯著相關的15種傳統計量指標。經主成分分析消除這15個指標間的相關性后,將其用作BP神經網絡CI值預測模型的輸入向量,同時采用"綜合性人文、社會科學"類的632個期刊的數據作為訓練樣本對網絡進行訓練。【結果/結論】使用訓練好的BP神經網絡對19種圖情領域核心期刊的CI值進行預測,結果顯示了較高的預測精度。該模型可用于影響力指數值的預測及期刊學科內排名的預估。

【Purpose/significance】This paper aims to discuss the correlation between academic journal Clout Index(CI)and the traditional journal evaluation indexes, and construct a prediction model for academic journal Clout Index.【Method/process】Nineteen core journals in library and information area are made as the research objects. The correlation betweenacademic journal Clout Index(CI) and twenty-two traditional journal evaluation indexes is analyzed by using SPSS 16.0.Fifteen traditional evaluation indexes which are significantly correlated with academic journal Clout Index are obtained.After eliminating the correlation among them by principal component analysis, these correlative indexes are made as theinput vector of BP neural network prediction model for CI value. Moreover, the data of six hundred and thirty-two journalsin comprehensive humanities and social sciences area become the training samples to train the network. 【Result/conclusion】 By using the trained BP neural network to predict the CI value of nineteen core journals in library andinformation area, the relative average error between the predicted value and the actual value can arrive at 2.18%, which is asatisfied precision. The proposed model can be used to estimate the academic journal Clout Index(CI) value and evaluatethe ranking of their journals.