Tendency For Items Research Articles

Personalised content recommendation based on field authorities in transparent computing

Transparent computing TC provides a large number of intelligent information services. Recommendation can help users rapidly locate their desired content. Collaborative filtering CF is a suitable recommending technology for TC, but data sparsity and noise problems have not been effectively solved. This study proposes a novel CF approach based on field authorities to achieve the genre tendency of items by mapping tags to genres and simulates a fine-grained word-of-mouth recommendation mode. Nearest neighbours are selected from sets of experienced users as field authorities in different genres, and weights are assigned to genres according to genre tendency. The method employed in this study can efficiently solve sparsity and noise problems and have high prediction accuracy. Experiments on MovieLens datasets show that the accuracy of this approach is significantly higher than that of traditional CF and expert CF in both mean absolute error and precision.

Separation of individual-level and cluster-level covariate effects in regression analysis of correlated data.

The focus of this paper is regression analysis of clustered data. Although the presence of intracluster correlation (the tendency for items within a cluster to respond alike) is typically viewed as an obstacle to good inference, the complex structure of clustered data offers significant analytic advantages over independent data. One key advantage is the ability to separate effects at the individual (or item-specific) level and the group (or cluster-specific) level. We review different approaches for the separation of individual-level and cluster-level effects on response, their appropriate interpretation and give recommendations for model fitting based on the intent of the data analyst. Unlike many earlier papers on this topic, we place particular emphasis on the interpretation of the cluster-level covariate effect. The main ideas of the paper are highlighted in an analysis of the relationship between birth weight and IQ using sibling data from a large birth cohort study.

A STUDY OF THE EFFECTS OF VARIATION OF SHORT‐TERM MEMORY LOAD, READING RESPONSE LENGTH, AND PROCESSING HIERARCHY ON TOEFL LISTENING COMPREHENSION ITEM PERFORMANCE

ABSTRACTThis study was conceived in response to criticisms of the current TOEFL listening comprehension test‐item format. Major areas of criticism have included speculation that listening as tested places too much burden on short‐term memory as opposed to comprehension, that a knowledge of reading is required in order to respond successfully, and that many items appear to require mere recall and matching of details rather than higher‐order processing skills. To address these criticisms in turn, a study was designed with 120 ESL learners and three listening tests (comprised of 144 total real and adapted TOEFL test items) to examine the characteristics of item functioning under conditions of stimulus repetition versus nonrepetition, variation of length of aural stimulus passage and of associated numbers of items, shorter versus longer reading response options, and higher versus lower level of processing skills required. Those item types and stimulus conditions that were found to associate with superior item functioning as indicated by estimates of item difficulty, item discriminability, internal consistency reliability, fit to a latent trait model, and convergent and discriminant validity were identified.Results suggested that, while repetition of the stimulus passage predictably tended to reduce item difficulty when control was made for concomitant influences, there was no consistent effect of stimulus passage repetition on item discrimination, Rasch model fit, or discriminant validity across difficulty level. However, there was a tendency for items in the no‐repetition condition to exhibit greater convergent and discriminant validity than items in the one‐repetition condition.Although passage length was confounded with numbers of items per passage and with comprehension hierarchy level, the test with passages of three‐sentence length tended to be more reliable than the test with passages of two‐sentence length, and the test with passages of two‐sentence length tended to be more reliable than the test with passages of one‐sentence length. Also, the test with the longest passages tended predictably to be slightly more difficult than the test with the shortest passages.Item response‐option length was significantly related to item difficulty and Rasch model fit in the direction that items with options that were shortened to about half current TOEFL response‐option length tended to be easier and to exhibit better fit than items with current longer options. Also, items with shortened options showed greater convergent and discriminant validity across levels of difficulty than did items with unshortened options. And, there was a near‐significant tendency for items with shortened options to exhibit better discrimination than items with unshortened options, when concomitant influences were controlled.Comprehension hierarchy level of items, as defined by the length of passage required to respond correctly, was not significantly related to item difficulty except through a complex option‐length‐by‐hierarchy‐ level interaction. However, hierarchy level was related to discrimination and Rasch model fit in the direction that items with lower level of processing (i.e., those that required comprehension of less stimulus text) showed better fit and discrimination than higher‐level items after concomitant influences were removed. Also, greater convergent and discriminant validity across difficulty levels was exhibited by lower‐level comprehension items than higher‐level items.It was concluded that tasks like those employed in TOEFL Listening Comprehension Section A would benefit from a shortening of current response‐option length, but that it was not beneficial to repeat stimulus passages, nor was it desirable to increase the proportion of items that depended on comprehension of greater rather than lesser amounts of text.

SOME FURTHER OBSERVATIONS ON THE MEASUREMENT OF CLUSTERING IN FREE RECALL

Properties of recently proposed measures of clustering in recall are considered. Applying some measures to computer‐generated sequences revealed that some of them are likely to give a spurious picture of differences in the tendency for items to cluster in sequences that also differ in length.