Extracting Features of Entertainment Products: A Guided Latent Dirichlet Allocation Approach Informed by the Psychology of Media Consumption:
Citations
321 citations
111 citations
103 citations
47 citations
Cites background from "Extracting Features of Entertainmen..."
...Although these text generators are not the firm's customers, they could nevertheless shed light on broad consumer cultural trends such as tastes in movies (Toubia et al., 2019) and music (Berger & Packard, 2018)....
[...]
30 citations
References
25,546 citations
12,650 citations
"Extracting Features of Entertainmen..." refers background in this paper
...Positive psychology is the branch of psychology that focuses on the achievement of a satisfactory life (for an introduction to positive psychology, see Seligman and Csikszentmihalyi 2000; Seligman et al. 2005)....
[...]
12,443 citations
Additional excerpts
...We consider replacing the bag-of-words variables created using LSA with guided LDA features (version 4), as both of these sets of features are based on some natural language processing of movie descriptions....
[...]
...Features Version 1 Version 2 Version 3 Version 4 Version 5 Intercept P P P P P Average critic rating P P P P Average user score P P P P Production budget P P P P Widest release P P P P Widest release2 P P P P Domestic box office P P P P MPAA rating P P P P Run time P P P P Competition P P P P Star power P P P P DVD release timing P P P P DVD sales rank P P P P Genres P P Content variables P P Semantic variables P P Bag-of-words variables from LSA P Guided LDA topic weights P P DIC 492.91 406.50 371.65 232.02 280.28 In-sample hit rate 62.09% 71.78% 76.30% 88.21% 85.08% Out-of-sample hit rate 61.67% 66.44% 67.94% 70.32% 71.19% Notes: Each column corresponds to one set of features....
[...]
...Replacing the unsupervised features based on LSA with guided LDA features results in significant improvement in in-sample and out-ofsample fit (version 4 vs. version 3)....
[...]
...It is a probabilistic version of LSA, thus enabling likelihoodbased inference....
[...]
...We find that for both the bagged-CART model based on Eliashberg, Hui, and Zhang (2007) and the kernelbased model from Eliashberg, Hui, and Zhang (2014), performance is improved when guided LDA features are included instead of the LSA bag-of-words variables....
[...]
9,873 citations
"Extracting Features of Entertainmen..." refers background or methods in this paper
...However, this approach also leads to a very large set of features, leading to a curse of dimensionality (Adomavicius and Tuzhilin 2005). are driven at least in part by the alignment of their psychological profile with the psychological themes featured in the product....
[...]
...However, this approach also leads to a very large set of features, leading to a curse of dimensionality (Adomavicius and Tuzhilin 2005)....
[...]
...Approaches for leveraging panel data in the media and entertainment industry have been classified into three categories (Adomavicius and Tuzhilin 2005): pure collaborative approaches, in which a user’s behavior is predicted on the basis of past behavior of similar users; content-based approaches,…...
[...]
9,583 citations
"Extracting Features of Entertainmen..." refers background or methods in this paper
...Popular collaborative approaches include variants of neighborhood-based collaborative filtering (CF; Breese, Heckerman, and Kadie 1998; Linden, Smith, and York 2003) and latent factor models (Koren, Bell, and Volinsky 2009)....
[...]
...In particular, CF is a very popular approach for predicting consumption at the individual level and recommending entertainment products to consumers (Koren, Bell, and Volinsky 2009; Linden, Smith, and York 2003)....
[...]