input.txt

Cross-domain adaptation for sentiment classification is the process of adapting a classifier that uses knowledge from one or multiple source domains and little data from the target domain to function with acceptable accuracy, precision, and recall on the target domain. One of the challenges facing cross-domain adaptation for sentiment classification is the limited availability of labeled samples in the target domain. In this paper, we introduce text generation in the target domain as a solution to provide a set of labeled data in the target domain then compare deep learning based text generators such as LSTM RNN and GRU RNN against Markov chain based text generators. We first use a rule-based classifier that utilizes knowledge from different source domains in labeling the unlabeled samples in the target domain (Kitchen Product reviews) then we have selected high confidence labeled samples for training LSTM RNN, GRU RNN and Markov chain based text generators. We have evaluated the deep learning based and Markov chain based text generators by measuring the fscores and accuracies of the end classifier when trained on the data generated from each of these models when tested on the kitchen benchmark test set.

Lexically constrained sentence generation allows the incorporation of prior knowledge such as lexical constraints into the output. This technique has been applied to machine translation, and dialog response generation. Previous work usually used Markov Chain Monte Carlo (MCMC) sampling to generate lexically constrained sentences, but they randomly determined the position to be edited and the action to be taken, resulting in many invalid refinements. To overcome this challenge, we used a classifier to instruct the MCMC-based models where and how to refine the candidate sentences. First, we developed two methods to create synthetic data on which the pre-trained model is fine-tuned to obtain a reliable classifier. Next, we proposed a two-step approach, “Predict and Revise”, for constrained sentence generation. During the predict step, we leveraged the classifier to compute the learned prior for the candidate sentence. During the revise step, we resorted to MCMC sampling to revise the candidate sentence by conducting a sampled action at a sampled position drawn from the learned prior. We compared our proposed models with many strong baselines on two tasks, generating sentences with lexical constraints and text infilling. Experimental results have demonstrated that our proposed model performs much better than the previous work in terms of sentence fluency and diversity. Our code, pre-trained models and Appendix are available at /~https://github.com/NLPCode/MCMCXLNet.

From ancient Egyptians to today’s internet users, people have always loved their cats.
In the U.S. alone, cats reign over about 45.3 million households. There are at least 45 domestic breeds, which differ widely in features such as coat color, tail length, hair texture, and temperament, according to the Cat Fancier’s Association.
The Maine Coon is the largest, with males reaching an average of 3.5 feet long. The smallest breed is the Singapura, native to Singapore, with adult females weighing as little as four pounds. One of the most unusual-looking cats is the Sphynx, a mostly hairless cat known for being robust and intelligent.
Like their big cat cousins, house cats are obligate carnivores, meaning they have to eat meat to stay healthy. Though they’ve been domesticated for thousands of years, these predators have maintained a strong hunting instinct, relying on stealth to stalk prey and attack with sharp claws and teeth. (Learn surprising things you never knew about your cat.)
As mostly nocturnal animals, cats have excellent vision and hearing, with ears that can turn like satellite dishes. Their reputation for having nine lives stems in part from their ability to navigate difficult environments, for example using their tail to balance and mostly land their lean, muscular bodies on all fours. Cushioning discs between vertebrae also give cat bodies exceptional flexibility and speed.
Kitty origins
People began to domesticate cats in the Fertile Crescent about 10,000 years ago, according to DNA research. Modern-day cats descended from a subspecies of African wildcat, Felis silvestris lybica, which today is the most common and widespread wildcat. (Read more about little-known small wildcats.)
Thousands of years ago, these wildcats were likely drawn to human settlements and their plentiful mice and food scraps. People realized these rodent catchers were helpful to have around, and eventually the two species began living together. Later, people began to bring felines aboard ships as they traveled the world. (Read about house cat ancestors’ remains found in Polish caves.)
Another, independent foray into cat domestication occurred in China about 5,000 years ago with another wildcat species the leopard cat. Since domestic cats today aren’t related to leopard cats, the harmony doesn’t seem to have lasted.
Reproduction
Females reach sexual maturity at just four months old and go into heat several times a year. Gestation lasts about 64 days, with an average litter size of four kittens. Young are usually weaned at two months old and grow rapidly, reaching adult size by the time they’re 10 months to a year old.
One litter of kittens can have multiple fathers, a phenomenon more likely in city cats due to crowding and lower aggression among males.