ABOUT SEARCHING AND SEARCH PRACTICIES

Introduction

Many search engines don’t index “stopwords”, words that are very common and have little meaning by themselves. The stopword list is often just the most frequent words in the langu…

There’s something new cooking in how Lucene scores text. Lucene just switched to something called BM25 in trunk. That means a new scoring formula for Solr and Elasticsearch.

What is Learning to Rank? Learning to rank ties machine learning into the search engine relevance.

In this blog, we explore what makes search distinct from other machine learning problems. How does one approach search ranking as a machine learning problem? We go through a couple of approaches that give you an intuition on how to evaluate a learning to rank method.

Our search relevancy engineer at OpenSource Connections (OSC), uses Quepid every day! Read on to learn how to implement a normalized discounted cumulative gain ranking quality scorer in Quepid

Welcome, dear reader, to my first OSC blog post. Let’s dive in! While search relevance is often equated with ensuring customers find what they need, that is only part...

Recently while experimenting with word2vec-based features with Learning to Rank, I was exploring using collocations to improve the accuracy of my embeddings. If you read the original word2vec paper...

105 falsehoods programmers believe about search, a complex field where competence is hard-won through training, practice, and experience.

This article gives an overview into the opportunities and challenges when applying advanced transformer models such as BERT to search.

We unwrap the mystery behind two popular search relevance metrics nDCG and ERR through visualization, and discuss their pros and cons.

Five years ago, I wrote an article called What is Search Relevance?. Back then, I had to shout to convince people to even notice whether search results were accurate...

Ensuring results are relevant is tricky but critical to a good search experience. Choosing an evaluationmetric to summarize the performance of the search engine can be equally challenging because...

Does anyone else remember the Segway? Segway was billed as the most revolutionary transportation innovation, well, ever. We would use it instead of cars; instead of biking and walking....

Search is a big deal — how else could we navigate the digital age where information overload is the status quo? But despite search being omnipresent, it can still...

What are the key roles in a Search Team and how can you find people to fill them both from within and without your organisation?

We will outline all the things you need to know about measuring the quality and effectiveness of your store’s site-search.

Today we’ll discuss KPIs for the actual product items that are presented to a buyer when they use your site-search.

How much money are you losing or gaining, depending on a site-search query? You need to be able to know if search is working, and you can with these KPIs!

Johannes Peter introduces query expansion in the first part of a series on fundamentals of query rewriting

Over the decades, information retrieval has been dominated by classical methods such as BM25. These lexical models are simple and effective yet vulnerable to vocabulary mismatch. With the introduction of pre-trained language models such as BERT and its relatives, deep retrieval models have achieved superior performance with their strong ability to capture semantic relationships. The downside is that training these deep models is computationally expensive, and suitable datasets are not always available for fine-tuning toward the target domain. While deep retrieval models work best on domains close to what they have been trained on, lexical models are comparatively robust across datasets and domains. This suggests that lexical and deep models can complement each other, retrieving different sets of relevant results. But how can these results effectively be combined? And can we learn something from language models to learn new indexing methods? This talk will delve into both these approaches and exemplify when they work well and not so well. We will take a closer look at different strategies to combine them to get the best of both, even in zero-shot cases where we don't have enough data to fine-tune the deep model. The Search track is presented by OpenSource Connections

Desingning Elasticsearch schemas and queries for E-Commerce websites

In the past decade, autocomplete has become a required feature for search engines. Today, searchers who type into a search box expect to…

The previous post focused on how to determine the best autocomplete suggestions based on query probability and query performance. In this…

Search queries are made up of characters. It’s easy to take characters for granted; indeed, people who have never implemented a search…

As we saw in the previous post, modeling search as a conversation makes it possible to overcome breakdowns in communication between the…

So far, we’ve focused on understanding searchers based entirely on the words they type into the search box. But search doesn’t occur in a…

In the previous post on query scoping, we discussed query tagging as a special case of named-entity recognition (NER). In this post, we’ll…