Natural language processing, aka NLP, is a broad and rapidly evolving segment of today's emerging digital technologies often generalized as Artificial Intelligence (AI). Wikipedia defines NLP as " a subfield of AI concerned with the interactions between computers and human (natural) languages, in particular how to program computers to process and analyze large amounts of natural language data." By harnessing NLP, AI can successfully imitate human speech, form naturally-flowing sentences and give human-to-machine interactions a personal touch. Perhaps because both concepts are related to words and languages, natural language processing is often confused with text mining: advanced analysis technique used to filter large amounts of research and extract the relevant info. Text mining is more than just a search tool; its algorithms can understand complex concepts and identify patterns and trends across million of articles to provide valuable information with unmistakable novelty factor.

Company seeks to enhance overall platform services and returns for shareholders. New York, NY – February 14, 2019 – Smartling, the premier software and language translation services company that established the category Enterprise Translation Cloud, today announces the company's 2019 merger and acquisition strategy. Smartling's Board of Directors and its CEO Jack Welde are looking to the marketplace for businesses that could strengthen the company's fast-growing software and language services position. The company posted record revenues and growth in 2018, including an increase in language services bookings of 112% year-over-year. With fourth quarter language services bookings having increased 211% year-over-year, the company sees a lot of potential for growth, specifically in this service line.

With the promising progress of deep neural networks, layer aggregation has been used to fuse information across layers in various fields, such as computer vision and machine translation. However, most of the previous methods combine layers in a static fashion in that their aggregation strategy is independent of specific hidden states. Inspired by recent progress on capsule networks, in this paper we propose to use routing-by-agreement strategies to aggregate layers dynamically. Specifically, the algorithm learns the probability of a part (individual layer representations) assigned to a whole (aggregated representations) in an iterative way and combines parts accordingly. We implement our algorithm on top of the state-of-the-art neural machine translation model TRANSFORMER and conduct experiments on the widely-used WMT14 English-German and WMT17 Chinese-English translation datasets. Experimental results across language pairs show that the proposed approach consistently outperforms the strong baseline model and a representative static aggregation model.

The ability to quickly and automatically translate anything you see using a web service is a powerful one, yet few expect much from it other than a tolerable version of a foreign article, menu or street sign. Shouldn't this amazing tool be put to better use? It can be, and a company called Lilt is quietly doing so -- but crucially, it isn't even trying to leave the human element behind. By combining the expertise of human translators with the speed and versatility of automated ones, you get the best of both worlds -- and potentially a major business opportunity. The problem with machine translation, when you really get down to it, is that it's bad. Sure, it won't mistake "tomato" for "potato," but it can't be trusted to do anything beyond accurately translate the literal meaning of a series of words.

Controlled generation of text is of high practical use. Recent efforts have made impressive progress in generating or editing sentences with given textual attributes (e.g., sentiment). This work studies a new practical setting of text content manipulation. Given a structured record, such as `(PLAYER: Lebron, POINTS: 20, ASSISTS: 10)', and a reference sentence, such as `Kobe easily dropped 30 points', we aim to generate a sentence that accurately describes the full content in the record, with the same writing style (e.g., wording, transitions) of the reference. The problem is unsupervised due to lack of parallel data in practice, and is challenging to minimally yet effectively manipulate the text (by rewriting/adding/deleting text portions) to ensure fidelity to the structured content. We derive a dataset from a basketball game report corpus as our testbed, and develop a neural method with unsupervised competing objectives and explicit content coverage constraints. Automatic and human evaluations show superiority of our approach over competitive methods including a strong rule-based baseline and prior approaches designed for style transfer.

Are you wondering how much to spend on Valentine's Day? There are so many ways to say "I love you." Those three little words take on multiple forms through a variety of languages, and even emojis. According to Google, the heart-eyes emoji () is the fourth most popular emoji worldwide, followed by the traditional heart () emoji. Google also says translations on its Translate service spike on Valentine's Day, and translations of "I love you" more than double on February 14.

Standard sequential generation methods assume a pre-specified generation order, such as text generation methods which generate words from left to right. In this work, we propose a framework for training models of text generation that operate in non-monotonic orders; the model directly learns good orders, without any additional annotation. Our framework operates by generating a word at an arbitrary position, and then recursively generating words to its left and then words to its right, yielding a binary tree. Learning is framed as imitation learning, including a coaching method which moves from imitating an oracle to reinforcing the policy's own preferences. Experimental results demonstrate that using the proposed method, it is possible to learn policies which generate text without pre-specifying a generation order, while achieving competitive performance with conventional left-to-right generation.

We consider the problem of making machine translation more robust to character-level variation at the source side, such as typos. Existing methods achieve greater coverage by applying subword models such as byte-pair encoding (BPE) and character-level encoders, but these methods are highly sensitive to spelling mistakes. We show how training on a mild amount of random synthetic noise can dramatically improve robustness to these variations, without diminishing performance on clean text. We focus on translation performance on natural noise, as captured by frequent corrections in Wikipedia edit logs, and show that robustness to such noise can be achieved using a balanced diet of simple synthetic noises at training time, without access to the natural noise data or distribution.

While machine translation has traditionally relied on large amounts of parallel corpora, a recent research line has managed to train both Neural Machine Translation (NMT) and Statistical Machine Translation (SMT) systems using monolingual corpora only. In this paper, we identify and address several deficiencies of existing unsupervised SMT approaches by exploiting subword information, developing a theoretically well founded unsupervised tuning method, and incorporating a joint refinement procedure. Moreover, we use our improved SMT system to initialize a dual NMT model, which is further fine-tuned through on-the-fly back-translation. Together, we obtain large improvements over the previous state-of-the-art in unsupervised machine translation. For instance, we get 22.5 BLEU points in English-to-German WMT 2014, 5.5 points more than the previous best unsupervised system, and 0.5 points more than the (supervised) shared task winner back in 2014.

Fine-tuning large pre-trained models is an effective transfer mechanism in NLP. However, in the presence of many downstream tasks, fine-tuning is parameter inefficient: an entire new model is required for every task. As an alternative, we propose transfer with adapter modules. Adapter modules yield a compact and extensible model; they add only a few trainable parameters per task, and new tasks can be added without revisiting previous ones. The parameters of the original network remain fixed, yielding a high degree of parameter sharing. To demonstrate adapter's effectiveness, we transfer the recently proposed BERT Transformer model to 26 diverse text classification tasks, including the GLUE benchmark. Adapters attain near state-of-the-art performance, whilst adding only a few parameters per task. On GLUE, we attain within 0.4% of the performance of full fine-tuning, adding only 3.6% parameters per task. By contrast, fine-tuning trains 100% of the parameters per task.