Improving information retrieval through correspondence analysis instead of latent semantic analysis

Abstract The initial dimensions extracted by latent semantic analysis (LSA) of a document-term matrix have been shown to mainly display marginal effects, which are irrelevant for information retrieval. To improve the performance LSA, usually elements raw weighted and weighting exponent singular values can be adjusted. An alternative retrieval technique that ignores effects is correspondence (CA). In this paper, LSA CA empirically compared. Moreover, it explored whether two weightings also CA. results four empirical datasets show always performs better than LSA. Weighting data CA; however, dependent improvement small. Adjusting value often improves extent depends on dataset number dimensions.