Concepts are used to solve the term-mismatch problem. However, we need an effective similarity measure between concepts. Word embedding presents a promising solution. We present in this study three approaches to build concepts vectors based on words vectors. We use a vector-based measure to estimate inter-concepts similarity. Our experiments show promising results. Furthermore, words and concepts become comparable. This could be used to improve conceptual indexing process.

Consumers rely on search results, social media and peer reviews to perform research and gather feedback on businesses they are considering visiting or products they're thinking of purchasing. And while it can be easy to turn a blind eye on the reviews your business receives, simply ignoring those review sites can be damaging to your bottom line. While this probably comes as no surprise, 95% of shoppers read online reviews before making a purchase. As they seek out peer reviews on brands or products that they're considering doing business with, they're looking for specific things. Consumers actually look for negative reviews to discover authentic feedback from real customers.

Traveloka is an online travel company that provides a one-stop platform for a range of ticketing services, including flights, accommodation, and attractions. As one of Southeast Asia's "unicorn" startups valued at over $1 billion, Traveloka is constantly searching for ways to improve their user experience. As part of this initiative, Traveloka has invested heavily in a number of artificial intelligence and machine learning projects. With an expanding list of 19 core product offerings, improving search capabilities was key to their continued growth. To do this, Traveloka built a search function to make it easy for users to browse the full range of products from a single search bar.

Neural ranking models for information retrieval (IR) use shallow or deep neural networks to rank search results in response to a query. Traditional learning to rank models employ supervised machine learning (ML) techniques--including neural networks--over hand-crafted IR features. By contrast, more recently proposed neural models learn representations of language from raw text that can bridge the gap between query and document vocabulary. Unlike classical learning to rank models and non-neural approaches to IR, these new ML techniques are data-hungry, requiring large scale training data before they can be deployed. This tutorial introduces basic concepts and intuitions behind neural IR models, and places them in the context of classical non-neural approaches to IR.

Over time, the technological pipeline has made some content marketers remain in chaos. With AI as the fast-evolving approach, the SEO pillars are however expected to become vital for future trades. The developing systems are expected to make work easier and reliable. So far, SEO has been taking considerable attention by marketers as it has become one of the primary tools for boosting sales. However, considerable attention should be given also to the integration of AI into SEO and how will it impact its functionality.

Transformer based models are creating tremendous impact in the space of NLP as they have proven to be effective in a wide range of tasks such as POS tagging, machine translation, named-entity recognition, and a series of text classification tasks. This year saw the introduction to a whole family of transformer-based language models such as BERT, Transformer-XL, and GPT-2, among others. Langauge models, in general, offer desirable properties that can be leveraged in a transfer learning setting where you train a model with large-scale data to learn the properties of language in an unsupervised setting. The resulting model and weights can then be fine-tuned and be applied in low-resourced regimes to address different NLP tasks. In particular, it's exciting to see the use of BERT in different domains such as text classification, text summarization, text generation, and information retrieval.

A natural progression in the field of computer vision following unprecedented progress in image classification tasks is towards video and video understanding, especially how it relates to identifying human subjects and activities. A number of datasets and benchmarks are being established in this area¹. In parallel, further progress is being made in 2D image related computer vision tasks such as fine-grained classification, image segmentation, 3D image construction, robot vision, scene flow estimation and human pose estimation. As part of my final Data Science project at Metis bootcamp, I've decided to marry these two parallel tracks -- video and human pose estimation in specific -- to create a content-based video search engine. Since applying 2D human pose estimation for video search is a novel idea with "no proof of concept", I have simplified my approach by selecting a single performer, fixed location single camera video footage of Salsa dance videos.

Code-mixing is the practice of alternating between two or more languages. Mostly observed in multilingual societies, its occurrence is increasing and therefore its importance. A major part of sentiment analysis research has been monolingual, and most of them perform poorly on code-mixed text. In this work, we introduce methods that use different kinds of multilingual and cross-lingual embeddings to efficiently transfer knowledge from monolingual text to code-mixed text for sentiment analysis of code-mixed text. Our methods can handle code-mixed text through a zero-shot learning. Our methods beat state-of-the-art on English-Spanish code-mixed sentiment analysis by absolute 3\% F1-score. We are able to achieve 0.58 F1-score (without parallel corpus) and 0.62 F1-score (with parallel corpus) on the same benchmark in a zero-shot way as compared to 0.68 F1-score in supervised settings. Our code is publicly available.

Google has informed us that you may see a spike in errors in the unparsable structured data report within Google Search Console. This is a bug in the reporting system and you do not need to worry. The issue happened between January 13, 2020 and January 16, 2020. Google wrote on the data anomalies page "Some users may see a spike in unparsable structured data errors. This was due to an internal misconfiguration that will be fixed soon, and can be ignored."

You can read more about USE in this paper. Let's first read the data. Because the quora dataset is huge and takes a lot of time, we will take only 1% of the data. This will take around 3 minutes for encoding and indexing. It will have 4000 questions.