Search

Topic model provides an easy means to analyze huge amount of untagged text as well as other data. A topic can be defined as a group of words that happen to occur together at a greater frequency. Topic models connects words that have similar kind of meanings and differentiate among words with different or multiple meanings. So, topic models in simple words are a set of algorithms that unveil the hidden thematic structure in a document collection. It allows us to order, search and outline different large records of texts. In this paper we present a survey on different topic modeling techniques which includes Latent Semantic Analysis (LSA), Probabilistic Latent Semantic Analysis (PLSA), and Latent Dirichlet Allocation (LDA) along with some of the extensions of LDA. The characteristics, limitations and applications of these topic modeling techniques are also studied and summarized.

A field experiment was conducted during 2013–14 at Research Farm, College of Horticulture, Mandsaur (Madhya Pradesh) to study the weed management practices in cumin. Twelve weed management treatments were laid in randomized block design with three replications. Weed free treatment recorded significant maximum growth and yield attributes of cumin followed by oxadiargyl at 0.06 kg/ha (PE) fb two hand weeding at 40 and 60 DAS. Similarly, significant maximum seed yield (595 kg/ha), straw yield (900 kg/ha) and harvest index (39.7%) was observed with oxadiargyl at 0.06 kg/ha as pre-emergence fb two hand weeding at 40 and 60 DAS. Maximum weed population was measured in case of weedy check, which was followed by hand weeding (40 DAS) at all the growth stages of cumin. It may be concluded that application of oxadiargyl at 0.06 kg/ha as pre-emergence fb two hand weeding at 40 and 60 DAS may be use for higher yield of cumin.