{"id":11859,"date":"2019-11-29T11:04:16","date_gmt":"2019-11-29T10:04:16","guid":{"rendered":"https:\/\/wordlift.io\/blog\/en\/?p=11859"},"modified":"2021-11-03T14:54:18","modified_gmt":"2021-11-03T13:54:18","slug":"write-meta-descriptions-bert","status":"publish","type":"post","link":"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/","title":{"rendered":"How to write meta descriptions using BERT"},"content":{"rendered":"\r\n\r\n\r\n\r\n\t\t\t\t\t<span style=\"font-weight: 400;\">If you are confused about meta descriptions in <a class=\"wl-entity-page-link\" title=\"search engine optimization\" href=\"https:\/\/wordlift.io\/blog\/en\/entity\/search-engine-optimization\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/search_engine_optimization;http:\/\/rdf.freebase.com\/ns\/m.019qb_;http:\/\/dbpedia.org\/resource\/Search_engine_optimization;http:\/\/de.dbpedia.org\/resource\/Suchmaschinenoptimierung;http:\/\/pt.dbpedia.org\/resource\/Otimiza\u00e7\u00e3o_para_motores_de_busca;http:\/\/lt.dbpedia.org\/resource\/Optimizavimas_paie\u0161kos_sistemoms;http:\/\/lv.dbpedia.org\/resource\/Mekl\u0113t\u0101jprogrammas_optimiz\u0101cija;http:\/\/hr.dbpedia.org\/resource\/Optimizacija_web_stranice;http:\/\/hu.dbpedia.org\/resource\/Keres\u0151optimaliz\u00e1l\u00e1s;http:\/\/uk.dbpedia.org\/resource\/\u041e\u043f\u0442\u0438\u043c\u0456\u0437\u0430\u0446\u0456\u044f_\u0434\u043b\u044f_\u043f\u043e\u0448\u0443\u043a\u043e\u0432\u0438\u0445_\u0441\u0438\u0441\u0442\u0435\u043c;http:\/\/id.dbpedia.org\/resource\/Optimisasi_mesin_pencari;http:\/\/en.dbpedia.org\/resource\/Search_engine_optimization;http:\/\/it.dbpedia.org\/resource\/Ottimizzazione_(motori_di_ricerca);http:\/\/es.dbpedia.org\/resource\/Posicionamiento_en_buscadores;http:\/\/et.dbpedia.org\/resource\/Otsingumootoritele_optimeerimine;http:\/\/ro.dbpedia.org\/resource\/Optimizare_pentru_motoare_de_c\u0103utare;http:\/\/nl.dbpedia.org\/resource\/Zoekmachineoptimalisatie;http:\/\/no.dbpedia.org\/resource\/S\u00f8kemotoroptimalisering;http:\/\/be.dbpedia.org\/resource\/\u041f\u043e\u0448\u0443\u043a\u0430\u0432\u0430\u044f_\u0430\u043f\u0442\u044b\u043c\u0456\u0437\u0430\u0446\u044b\u044f;http:\/\/ru.dbpedia.org\/resource\/\u041f\u043e\u0438\u0441\u043a\u043e\u0432\u0430\u044f_\u043e\u043f\u0442\u0438\u043c\u0438\u0437\u0430\u0446\u0438\u044f;http:\/\/fi.dbpedia.org\/resource\/Hakukoneoptimointi;http:\/\/bg.dbpedia.org\/resource\/\u041e\u043f\u0442\u0438\u043c\u0438\u0437\u0430\u0446\u0438\u044f_\u0437\u0430_\u0442\u044a\u0440\u0441\u0430\u0447\u043a\u0438;http:\/\/fr.dbpedia.org\/resource\/Optimisation_pour_les_moteurs_de_recherche;http:\/\/sk.dbpedia.org\/resource\/Optimaliz\u00e1cia_pre_vyh\u013ead\u00e1va\u010de;http:\/\/sl.dbpedia.org\/resource\/Optimizacija_spletnih_strani;http:\/\/ca.dbpedia.org\/resource\/Optimitzaci\u00f3_per_a_motors_de_cerca;http:\/\/sq.dbpedia.org\/resource\/SEO;http:\/\/sr.dbpedia.org\/resource\/SEO_optimalizacija_veb-sajta;http:\/\/sv.dbpedia.org\/resource\/S\u00f6kmotoroptimering;http:\/\/cs.dbpedia.org\/resource\/Search_Engine_Optimization;http:\/\/pl.dbpedia.org\/resource\/Optymalizacja_dla_wyszukiwarek_internetowych;http:\/\/da.dbpedia.org\/resource\/S\u00f8gemaskineoptimering;http:\/\/tr.dbpedia.org\/resource\/Arama_motoru_optimizasyonu;http:\/\/data.wordlift.io\/wl0216\/entity\/search_engine_optimization\" >SEO<\/a>, why they are important and how to nail it with <\/span><i><span style=\"font-weight: 400;\">the help of <\/span><\/i><a href=\"https:\/\/wordlift.io\/blog\/en\/entity\/artificial-intelligence\/\"><i><span style=\"font-weight: 400;\">artificial intelligence<\/span><\/i><\/a><span style=\"font-weight: 400;\">, this article is for you.\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">If you are eager to start experimenting with an <a class=\"wl-entity-page-link\" title=\"Google\" href=\"https:\/\/wordlift.io\/blog\/en\/entity\/artificial-intelligence\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/artificial_intelligence;http:\/\/rdf.freebase.com\/ns\/m.0mkz;http:\/\/dbpedia.org\/resource\/Artificial_intelligence;http:\/\/data.wordlift.io\/wl0216\/entity\/artificial_intelligence_2;http:\/\/data.wordlift.io\/wl0216\/entity\/artificial_intelligence;http:\/\/data.wordlift.io\/wl0216\/entity\/artificial_intelligence_2;http:\/\/pt.dbpedia.org\/resource\/Intelig\u00eancia_artificial;http:\/\/hr.dbpedia.org\/resource\/Umjetna_inteligencija;http:\/\/hu.dbpedia.org\/resource\/Mesters\u00e9ges_intelligencia;http:\/\/id.dbpedia.org\/resource\/Kecerdasan_buatan;http:\/\/is.dbpedia.org\/resource\/Gervigreind;http:\/\/it.dbpedia.org\/resource\/Intelligenza_artificiale;http:\/\/ro.dbpedia.org\/resource\/Inteligen\u021b\u0103_artificial\u0103;http:\/\/be.dbpedia.org\/resource\/\u0428\u0442\u0443\u0447\u043d\u044b_\u0456\u043d\u0442\u044d\u043b\u0435\u043a\u0442;http:\/\/ru.dbpedia.org\/resource\/\u0418\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u044b\u0439_\u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442;http:\/\/bg.dbpedia.org\/resource\/\u0418\u0437\u043a\u0443\u0441\u0442\u0432\u0435\u043d_\u0438\u043d\u0442\u0435\u043b\u0435\u043a\u0442;http:\/\/sk.dbpedia.org\/resource\/Umel\u00e1_inteligencia;http:\/\/sl.dbpedia.org\/resource\/Umetna_inteligenca;http:\/\/ca.dbpedia.org\/resource\/Intel\u00b7lig\u00e8ncia_artificial;http:\/\/sq.dbpedia.org\/resource\/Inteligjenca_artificiale;http:\/\/sr.dbpedia.org\/resource\/\u0412\u0458\u0435\u0448\u0442\u0430\u0447\u043a\u0430_\u0438\u043d\u0442\u0435\u043b\u0438\u0433\u0435\u043d\u0446\u0438\u0458\u0430;http:\/\/sv.dbpedia.org\/resource\/Artificiell_intelligens;http:\/\/cs.dbpedia.org\/resource\/Um\u011bl\u00e1_inteligence;http:\/\/da.dbpedia.org\/resource\/Kunstig_intelligens;http:\/\/tr.dbpedia.org\/resource\/Yapay_zek\u00e2;http:\/\/de.dbpedia.org\/resource\/K\u00fcnstliche_Intelligenz;http:\/\/lt.dbpedia.org\/resource\/Dirbtinis_intelektas;http:\/\/lv.dbpedia.org\/resource\/M\u0101ksl\u012bgais_intelekts;http:\/\/uk.dbpedia.org\/resource\/\u0428\u0442\u0443\u0447\u043d\u0438\u0439_\u0456\u043d\u0442\u0435\u043b\u0435\u043a\u0442;http:\/\/en.dbpedia.org\/resource\/Artificial_intelligence;http:\/\/es.dbpedia.org\/resource\/Inteligencia_artificial;http:\/\/et.dbpedia.org\/resource\/Tehisintellekt;http:\/\/nl.dbpedia.org\/resource\/Kunstmatige_intelligentie;http:\/\/no.dbpedia.org\/resource\/Kunstig_intelligens;http:\/\/fi.dbpedia.org\/resource\/Teko\u00e4ly;http:\/\/fr.dbpedia.org\/resource\/Intelligence_artificielle;http:\/\/pl.dbpedia.org\/resource\/Sztuczna_inteligencja\" >AI<\/a>-writer, read the full article. At the end, I will give you a script to help you write meta descriptions on scale using <a class=\"wl-entity-page-link\"  href=\"https:\/\/wordlift.io\/blog\/en\/entity\/bert\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/bert;http:\/\/www.wikidata.org\/entity\/Q63612096\" >BERT<\/a>: <a class=\"wl-entity-page-link\" title=\"Google\u2019s\" href=\"https:\/\/wordlift.io\/blog\/en\/entity\/google\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/google;http:\/\/rdf.freebase.com\/ns\/m.045c7b;http:\/\/yago-knowledge.org\/resource\/Google;http:\/\/dbpedia.org\/resource\/Google;http:\/\/pt.dbpedia.org\/resource\/Google;http:\/\/hr.dbpedia.org\/resource\/Google_(tvrtka);http:\/\/hu.dbpedia.org\/resource\/Google_Inc.;http:\/\/id.dbpedia.org\/resource\/Google;http:\/\/is.dbpedia.org\/resource\/Google;http:\/\/it.dbpedia.org\/resource\/Google_Inc.;http:\/\/ro.dbpedia.org\/resource\/Google;http:\/\/ru.dbpedia.org\/resource\/Google_(\u043a\u043e\u043c\u043f\u0430\u043d\u0438\u044f);http:\/\/be.dbpedia.org\/resource\/Google;http:\/\/bg.dbpedia.org\/resource\/\u0413\u0443\u0433\u044a\u043b;http:\/\/sk.dbpedia.org\/resource\/Google;http:\/\/sl.dbpedia.org\/resource\/Google;http:\/\/ca.dbpedia.org\/resource\/Google;http:\/\/sq.dbpedia.org\/resource\/Google;http:\/\/sr.dbpedia.org\/resource\/\u0413\u0443\u0433\u043b;http:\/\/sv.dbpedia.org\/resource\/Google;http:\/\/cs.dbpedia.org\/resource\/Google;http:\/\/da.dbpedia.org\/resource\/Google;http:\/\/tr.dbpedia.org\/resource\/Google;http:\/\/de.dbpedia.org\/resource\/Google_Inc.;http:\/\/lt.dbpedia.org\/resource\/Google;http:\/\/lv.dbpedia.org\/resource\/Google;http:\/\/uk.dbpedia.org\/resource\/Google;http:\/\/en.dbpedia.org\/resource\/Google;http:\/\/es.dbpedia.org\/resource\/Google;http:\/\/et.dbpedia.org\/resource\/Google;http:\/\/nl.dbpedia.org\/resource\/Google_Inc.;http:\/\/no.dbpedia.org\/resource\/Google;http:\/\/fi.dbpedia.org\/resource\/Google;http:\/\/fr.dbpedia.org\/resource\/Google;http:\/\/pl.dbpedia.org\/resource\/Google\" >Google<\/a>\u2019s pre-trained, unsupervised language model that has recently gained great momentum in the SEO community after both, <\/span><a href=\"https:\/\/www.blog.google\/products\/search\/search-language-understanding-bert\/\"><span style=\"font-weight: 400;\">Google<\/span><\/a><span style=\"font-weight: 400;\"> and <\/span><a href=\"https:\/\/azure.microsoft.com\/en-us\/blog\/bing-delivers-its-largest-improvement-in-search-experience-using-azure-gpus\/\"><span style=\"font-weight: 400;\">BING<\/span><\/a><span style=\"font-weight: 400;\"> announced that they use it for providing more useful results.\u00a0\u00a0\u00a0\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">I used to underestimate the importance of meta descriptions myself: after all Google will use it only on 35.9% of the cases (according to <\/span><span style=\"font-weight: 400;\">a <\/span><a href=\"https:\/\/moz.com\/blog\/how-long-should-your-meta-description-be-2018\"><span style=\"font-weight: 400;\">Moz analysis<\/span><\/a><span style=\"font-weight: 400;\"> from last year by the illustrious <\/span><a href=\"http:\/\/twitter.com\/dr_pete\"><span style=\"font-weight: 400;\">@dr_pete<\/span><\/a><span style=\"font-weight: 400;\">). In reality, these brief snippets of text, <\/span><b>greatly help to entice more users to your website <\/b><span style=\"font-weight: 400;\">and, indirectly, might even <\/span><b>influence your ranking<\/b><span style=\"font-weight: 400;\"> thanks to higher<\/span><b> click-through-rate (CTR)<\/b><span style=\"font-weight: 400;\">.\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">While Google can overrule the meta descriptions added in the <a class=\"wl-entity-page-link\"  href=\"https:\/\/wordlift.io\/blog\/en\/entity\/html\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/html;http:\/\/rdf.freebase.com\/ns\/m.03g20;http:\/\/yago-knowledge.org\/resource\/HTML;http:\/\/dbpedia.org\/resource\/HTML\" >HTML<\/a> of your pages, if you properly align:<\/span>\r\n<ol>\r\n \t<li style=\"font-weight: 400;\"><b><i>the main intent<\/i><\/b><span style=\"font-weight: 400;\"> of the user (the query you are targeting),\u00a0<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><b><i>the title of the page<\/i><\/b> <span style=\"font-weight: 400;\">and<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><b><i>the meta description<\/i><\/b><b>\u00a0<\/b><\/li>\r\n<\/ol>\r\n<span style=\"font-weight: 400;\">There are many possibilities to improve the CTR on Google\u2019s result pages. In the course of this article we will investigate the following aspects and, since it\u2019s a long article, feel free to jump to the section that interests you the most &#8212; code is available at the end.<\/span>\r\n<h2><span style=\"font-weight: 400;\">What are meta descriptions?<\/span><\/h2>\r\n<span style=\"font-weight: 400;\">As usual I tend to \u201cask\u201d\u00a0 \u201cexperts\u201d online a definition to get started, and with <\/span><a href=\"https:\/\/www.google.com\/search?q=woorank+what+is+a+meta+description&amp;oq=woorank+what&amp;aqs=chrome.0.69i59j69i57j35i39j69i59.3990j0j7&amp;sourceid=chrome&amp;ie=UTF-8\"><span style=\"font-weight: 400;\">a simple query<\/span><\/a><span style=\"font-weight: 400;\"> on Google, we can get this definition from our friends at WooRank:<\/span>\r\n\r\n<b>Meta descriptions<\/b><span style=\"font-weight: 400;\"><\/span> are HTML <b>tags<\/b><span style=\"font-weight: 400;\"> that appear in the head section of a web page. The content within the <\/span><b>tag<\/b><span style=\"font-weight: 400;\"> provides a <\/span><b>description<\/b><span style=\"font-weight: 400;\"><\/span> of what the page and its content are about. In the context of SEO, <b>meta descriptions<\/b><span style=\"font-weight: 400;\"><\/span> should be around 160 characters long.<span style=\"font-weight: 400;\"><\/span>\r\n\r\n<span style=\"font-weight: 400;\"><img decoding=\"async\" class=\"alignnone size-large wp-image-11861\" src=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1-1024x645.png\" alt=\"meta description definition\" width=\"1024\" height=\"645\" srcset=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1-1024x645.png 1024w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1-300x189.png 300w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1-768x484.png 768w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1-150x94.png 150w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1.png 1324w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/span>\r\n\r\n<span style=\"font-weight: 400;\">Here\u2019s an example of what a meta description usually looks like (from that same article):<\/span>\r\n\r\n<img decoding=\"async\" class=\"alignnone size-large wp-image-11862\" src=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1.1-1024x208.png\" alt=\"meta description example\" width=\"1024\" height=\"208\" srcset=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1.1-1024x208.png 1024w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1.1-300x61.png 300w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1.1-768x156.png 768w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1.1-150x30.png 150w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert1.1.png 1232w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/>\r\n<h2><span style=\"font-weight: 400;\">How long should your meta description be?<\/span><\/h2>\r\n<span style=\"font-weight: 400;\">We want to be, as with any other content on our site, authentic, conversational and user-friendly. Having said that, in 2020, you will want to stick to the <\/span><b>155-160 characters limit<\/b><span style=\"font-weight: 400;\"><\/span> (this corresponds to <b>920 pixels<\/b><span style=\"font-weight: 400;\"><\/span>). We also want to keep in mind that the \u201coptimal\u201d length might change based on the query of the user. This means that you should really <b>do your best in the first 120 characters<\/b><span style=\"font-weight: 400;\"><\/span> and think in terms of <b>creating a meaningful chain by linking the query, the title tag and the meta description<\/b><span style=\"font-weight: 400;\"><\/span>. In some cases, within this chain it is also very important to consider the role of the <b>breadcrumbs<\/b><span style=\"font-weight: 400;\">. As in the example above from WooRank I can quickly see that the definition is coming from an educational page of their site: this fits very well with my information request.\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/span>\r\n<h2><span style=\"font-weight: 400;\">What meta descriptions should we focus on?<\/span><\/h2>\r\n<b>SEO is a process<\/b><span style=\"font-weight: 400;\"><\/span>: we need to set our goals, analyze the data we\u2019re starting with, improve our content, and measure the results. There is no point in looking at a large website and saying, I need to <a class=\"wl-entity-page-link\" title=\"Writing\" href=\"https:\/\/wordlift.io\/blog\/en\/entity\/writing\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/writing;http:\/\/rdf.freebase.com\/ns\/m.081rb;http:\/\/dbpedia.org\/resource\/Writing\" >write<\/a> a gazillion of meta descriptions since they are all missing. It would simply be a waste of time.\r\n\r\n<span style=\"font-weight: 400;\">Besides the fact that in some cases &#8211; <\/span><b>we might decide not to add a meta description at all.<\/b><span style=\"font-weight: 400;\"> For example, when a page covers different queries and the text is already well structured we might leave it to Google to craft the best snippet for each super query (they are super good at it ?). We need to <\/span><b>look at the critical pages<\/b><span style=\"font-weight: 400;\"><\/span> we have &#8211; let\u2019s not forget that writing a good meta description is just like writing an ad copy \u2014 driving clicks is not a trivial game.\r\n\r\n<span style=\"font-weight: 400;\">As a rule of thumb I prefer to focus my attention on:\u00a0<\/span>\r\n<ul>\r\n \t<li style=\"font-weight: 400;\"><b>Pages that are already ranking on Google (<\/b><b>position &gt; 0<\/b><b>)<\/b><span style=\"font-weight: 400;\">; adding a meta description to a page that is not ranking will not make a difference.<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><b>Pages that are not in the top 3 positions: <\/b><span style=\"font-weight: 400;\">if they are already highly ranked, unless I can see some real opportunities &#8211; I prefer to leave them as they are.<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><b>Pages that have a business value<\/b><span style=\"font-weight: 400;\"><\/span>: on the wordlift website (the company I work for), there is no point in adding meta descriptions to landing pages that have no organic potential. I would rather prefer to focus on content from our blog. This varies of course but is very important to understand what type of pages I want to focus on.<\/li>\r\n<\/ul>\r\n<span style=\"font-weight: 400;\">This criteria can be useful, especially if you plan to <\/span><a href=\"https:\/\/www.woorank.com\/en\/features\/site-crawl\"><span style=\"font-weight: 400;\">programmatically crawl our website<\/span><\/a><span style=\"font-weight: 400;\"> and choose where to focus our attention using crawl data. Keep on reading and we\u2019ll get there, I promise.\u00a0<\/span>\r\n<h2><span style=\"font-weight: 400;\">A quick introduction to single-document text summarization<\/span><\/h2>\r\n<span style=\"font-weight: 400;\">Automatic text summarization is a challenging <\/span><a href=\"https:\/\/wordlift.io\/blog\/en\/entity\/natural-language-processing\/\"><span style=\"font-weight: 400;\">NLP<\/span><\/a><span style=\"font-weight: 400;\"> task to provide a short and possibly accurate summary of a long text. While, with the growing amount of online content, the need for understanding and summarizing content is very high. In pure technological terms, the challenge for creating well formed summaries is huge and results are, most of the time, still far from being perfect (or human-level).<\/span>\r\n\r\n<span style=\"font-weight: 400;\">The first research work on automatic text summarization goes back to 50 years ago and various techniques. Since then, they have been used to extract relevant content from unstructured text.\u00a0\u00a0<\/span>\r\n\r\n<i><span style=\"font-weight: 400;\">\u201cThe different dimensions of text summarization can be generally categorized based on its input type (single or multi document), purpose (generic, domain specific, or query-based) and output type (extractive or abstractive).\u201d<\/span><\/i>\r\n\r\n<span style=\"font-weight: 400;\">\u2014 <\/span><a href=\"http:\/\/thescipub.com\/PDF\/jcssp.2016.178.190.pdf\"><span style=\"font-weight: 400;\">A Review on Automatic Text Summarization Approaches<\/span><\/a><span style=\"font-weight: 400;\">, 2016.<\/span>\r\n<h2><span style=\"font-weight: 400;\">Extractive vs Abstractrive\u00a0<\/span><\/h2>\r\n<span style=\"font-weight: 400;\">Let\u2019s have a quick look at the different methods we have for compressing a web page.\u00a0<\/span>\r\n\r\n<img decoding=\"async\" class=\"alignnone size-large wp-image-11863\" src=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert2-1024x454.png\" alt=\"Extractive and Abstractive Summarization\" width=\"1024\" height=\"454\" srcset=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert2-1024x454.png 1024w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert2-300x133.png 300w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert2-768x340.png 768w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert2-150x66.png 150w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert2.png 1214w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/>\r\n\r\n<span style=\"font-weight: 400;\">\u201cExtractive summarization methods work by identifying important sections of the text and generating them verbatim; [\u2026] abstractive summarization methods aim at producing important material in a new way. In other words, they interpret and examine the text using advanced natural language techniques in order to generate a new shorter text that conveys the most critical information from the original text\u201d<\/span>\r\n\r\n<span style=\"font-weight: 400;\">\u2014 <\/span><a href=\"https:\/\/arxiv.org\/abs\/1707.02268\"><span style=\"font-weight: 400;\">Text Summarization Techniques: A Brief Survey<\/span><\/a><span style=\"font-weight: 400;\">, 2017.<\/span>\r\n\r\n<span style=\"font-weight: 400;\">With simple words with extractive summarization we will use an algorithm to select and combine the most relevant sentences in a document. Using abstractive summarization methods, we will use sophisticated <a class=\"wl-entity-page-link\" title=\"Natural language processing\" href=\"https:\/\/wordlift.io\/blog\/en\/entity\/natural-language-processing\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/natural_language_processing;http:\/\/rdf.freebase.com\/ns\/m.05flf;http:\/\/dbpedia.org\/resource\/Natural_language_processing;http:\/\/be.dbpedia.org\/resource\/\u0410\u043f\u0440\u0430\u0446\u043e\u045e\u043a\u0430_\u043d\u0430\u0442\u0443\u0440\u0430\u043b\u044c\u043d\u0430\u0439_\u043c\u043e\u0432\u044b;http:\/\/ru.dbpedia.org\/resource\/\u041e\u0431\u0440\u0430\u0431\u043e\u0442\u043a\u0430_\u0435\u0441\u0442\u0435\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e_\u044f\u0437\u044b\u043a\u0430;http:\/\/pt.dbpedia.org\/resource\/Processamento_de_linguagem_natural;http:\/\/bg.dbpedia.org\/resource\/\u041e\u0431\u0440\u0430\u0431\u043e\u0442\u043a\u0430_\u043d\u0430_\u0435\u0441\u0442\u0435\u0441\u0442\u0432\u0435\u043d_\u0435\u0437\u0438\u043a;http:\/\/lt.dbpedia.org\/resource\/Nat\u016bralios_kalbos_apdorojimas;http:\/\/fr.dbpedia.org\/resource\/Traitement_automatique_du_langage_naturel;http:\/\/uk.dbpedia.org\/resource\/\u041e\u0431\u0440\u043e\u0431\u043a\u0430_\u043f\u0440\u0438\u0440\u043e\u0434\u043d\u043e\u0457_\u043c\u043e\u0432\u0438;http:\/\/id.dbpedia.org\/resource\/Pemrosesan_bahasa_alami;http:\/\/ca.dbpedia.org\/resource\/Processament_de_llenguatge_natural;http:\/\/sr.dbpedia.org\/resource\/Obrada_prirodnih_jezika;http:\/\/en.dbpedia.org\/resource\/Natural_language_processing;http:\/\/is.dbpedia.org\/resource\/M\u00e1lgreining;http:\/\/it.dbpedia.org\/resource\/Elaborazione_del_linguaggio_naturale;http:\/\/es.dbpedia.org\/resource\/Procesamiento_de_lenguajes_naturales;http:\/\/cs.dbpedia.org\/resource\/Zpracov\u00e1n\u00ed_p\u0159irozen\u00e9ho_jazyka;http:\/\/pl.dbpedia.org\/resource\/Przetwarzanie_j\u0119zyka_naturalnego;http:\/\/ro.dbpedia.org\/resource\/Prelucrarea_limbajului_natural;http:\/\/da.dbpedia.org\/resource\/Sprogteknologi;http:\/\/tr.dbpedia.org\/resource\/Do\u011fal_dil_i\u015fleme\" >NLP<\/a> techniques (i.e. deep neural networks) to read and understand a document in order to generate novel sentences.\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">In extractive methods a document can be seen as a graph where each sentence is a node and the relationships between these sentences are weighted edges. These edges can be computed by analyzing the similarity between the word-sets from each sentence. We can then use an algorithm like Page Rank (we will call it Text Rank in this context) to extract <\/span><i><span style=\"font-weight: 400;\">the most central sentences<\/span><\/i><span style=\"font-weight: 400;\"> in our document-graph.<\/span>\r\n\r\n<em>\u00a0<img decoding=\"async\" class=\"alignnone size-large wp-image-11864\" src=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert3-1024x525.png\" alt=\"Text Rank algorithm\" width=\"1024\" height=\"525\" srcset=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert3-1024x525.png 1024w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert3-300x154.png 300w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert3-768x394.png 768w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert3-150x77.png 150w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert3.png 1174w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/em>\r\n<h2><span style=\"font-weight: 400;\">The carbon footprint of NLP and why I prefer extractive methods to create meta descriptions<\/span><\/h2>\r\n<span style=\"font-weight: 400;\">In a recent <\/span><a href=\"http:\/\/arxiv.org\/abs\/1906.02243\"><span style=\"font-weight: 400;\">study<\/span><\/a><span style=\"font-weight: 400;\">, researchers at the University of Massachusetts, Amherst, performed a life cycle assessment for training several common large AI models with focus on language models and NLP tasks. They found that <\/span><b>training a complex language model can emit<\/b> <b>five times the lifetime emissions of the average American car (including whatever is required to manufacture the car itself!).<\/b><em>\u00a0<\/em>\r\n\r\n<span style=\"font-weight: 400;\">While automation is key we don\u2019t want to contribute to the pollution of\u00a0 our planet by misusing the technology we have. In principle, using abstract methods and deep learning techniques offers a higher degree of control when compressing articles into 30-60 word paragraphs but, considering our end goal (enticing more clicks from organic search), we can probably find a good compromise without spending too many computational (and environmental) resources. I know it sounds a bit na\u00efve but&#8230;it is not and we want to be sustainable and efficient in everything we do.<\/span>\r\n<h2><span style=\"font-weight: 400;\">What is BERT?<\/span><\/h2>\r\n<h3><span style=\"font-weight: 400;\">BERT: The Mighty Transformer\u00a0<\/span><\/h3>\r\n<span style=\"font-weight: 400;\">Now, provided the fact that a significant amount of energy has been already spent to train BERT (1,507 kWh according to the paper mentioned above), I decided it was worth testing it for running <\/span><b>extractive summarization<\/b><span style=\"font-weight: 400;\">.\u00a0<\/span>\r\n\r\n<em>\u00a0<img decoding=\"async\" class=\"alignnone size-full wp-image-11865\" src=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert4.png\" alt=\"Bert from Sesame Street\" width=\"400\" height=\"400\" srcset=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert4.png 400w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert4-300x300.png 300w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert4-150x150.png 150w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert4-96x96.png 96w\" sizes=\"(max-width: 400px) 100vw, 400px\" \/><\/em>\r\n\r\n<span style=\"font-weight: 400;\">I have also to admit that It has been quite some time since I entertained myself with automatic text-summarization of online content and I have experimented with a lot of different methods before getting into BERT.\u00a0\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">BERT is a pre-trained unsupervised natural language processing model created by Google and released as an <a class=\"wl-entity-page-link\" title=\"open-source\" href=\"https:\/\/wordlift.io\/blog\/en\/entity\/open-source\/\" data-id=\"http:\/\/data.wordlift.io\/wl0216\/entity\/open_source;http:\/\/rdf.freebase.com\/ns\/m.02wtqd4;http:\/\/yago-knowledge.org\/resource\/Open_source;http:\/\/dbpedia.org\/resource\/Open_source;http:\/\/de.dbpedia.org\/resource\/Open_Source;http:\/\/pt.dbpedia.org\/resource\/C\u00f3digo_aberto;http:\/\/lt.dbpedia.org\/resource\/Atvirasis_kodas;http:\/\/hr.dbpedia.org\/resource\/Otvoreni_kod;http:\/\/lv.dbpedia.org\/resource\/Atv\u0113rtais_pirmkods;http:\/\/uk.dbpedia.org\/resource\/\u041f\u043e\u043b\u0456\u0442\u0438\u043a\u0430_\u0432\u0456\u0434\u043a\u0440\u0438\u0442\u043e\u0433\u043e_\u043a\u043e\u0434\u0443;http:\/\/id.dbpedia.org\/resource\/Sumber_terbuka;http:\/\/en.dbpedia.org\/resource\/Open_source;http:\/\/is.dbpedia.org\/resource\/Opinn_hugb\u00fana\u00f0ur;http:\/\/it.dbpedia.org\/resource\/Open_source;http:\/\/es.dbpedia.org\/resource\/C\u00f3digo_abierto;http:\/\/et.dbpedia.org\/resource\/Avatud_l\u00e4htekood;http:\/\/ro.dbpedia.org\/resource\/Surs\u0103_deschis\u0103;http:\/\/nl.dbpedia.org\/resource\/Open_source;http:\/\/no.dbpedia.org\/resource\/\u00c5pen_kildekode;http:\/\/fi.dbpedia.org\/resource\/Avoin_l\u00e4hdekoodi;http:\/\/bg.dbpedia.org\/resource\/\u041e\u0442\u0432\u043e\u0440\u0435\u043d_\u043a\u043e\u0434;http:\/\/fr.dbpedia.org\/resource\/Open_source;http:\/\/sk.dbpedia.org\/resource\/Open_source;http:\/\/sl.dbpedia.org\/resource\/Odprta_koda;http:\/\/ca.dbpedia.org\/resource\/Codi_obert;http:\/\/sq.dbpedia.org\/resource\/Open_source;http:\/\/sr.dbpedia.org\/resource\/\u041e\u0442\u0432\u043e\u0440\u0435\u043d\u0438_\u043a\u043e\u0434;http:\/\/sv.dbpedia.org\/resource\/\u00d6ppen_k\u00e4llkod;http:\/\/da.dbpedia.org\/resource\/Open_source;http:\/\/tr.dbpedia.org\/resource\/A\u00e7\u0131k_kaynak\" >open source<\/a> program (yay!) that does magic on 11 of the most common NLP tasks.<\/span>\r\n\r\n<span style=\"font-weight: 400;\">BERTSUM, is a variant of BERT, designed for extractive summarization that is now state-of-the-art (<\/span><a href=\"https:\/\/arxiv.org\/pdf\/1903.10318v2.pdf\"><span style=\"font-weight: 400;\">here<\/span><\/a><span style=\"font-weight: 400;\"> you can find the paper behind it).\u00a0<\/span>\r\n\r\n<a href=\"https:\/\/github.com\/dmmiller612\"><span style=\"font-weight: 400;\">Derek Miller<\/span><\/a><span style=\"font-weight: 400;\">, leveraging on these progresses, has done a terrific work for bringing this technology to the masses (myself included) by creating a super sleek and easy-to-use Python library that we can use to experiment BERT-powered extractive text summarization at scale. A big thank you also goes to the HuggingFace team since Derek\u2019s tool uses their Pytorch transformers library ?.\u00a0<\/span>\r\n<h2><span style=\"font-weight: 400;\">Long live AI, let\u2019s scale the generation of meta descriptions with our adorable robot [<\/span><a href=\"https:\/\/colab.research.google.com\/drive\/1TeASgjfCGiZY7VjxcQmHIHHjy5L9Z-j1\"><span style=\"font-weight: 400;\">CODE IS HERE<\/span><\/a><span style=\"font-weight: 400;\">]<\/span><\/h2>\r\n<span style=\"font-weight: 400;\">So here is how everything works in the code linked to this article.\u00a0<\/span>\r\n\r\n<img decoding=\"async\" class=\"alignnone size-large wp-image-11866\" src=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert5-1024x289.png\" alt=\"Infograph of our AI\" width=\"1024\" height=\"289\" srcset=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert5-1024x289.png 1024w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert5-300x85.png 300w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert5-768x217.png 768w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert5-1536x434.png 1536w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert5-150x42.png 150w, https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert5.png 1580w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/>\r\n<ol>\r\n \t<li style=\"font-weight: 400;\"><span style=\"font-weight: 400;\">We start with a CSV that I generated using the WooRank\u2019s crawler (here you can tweak the code and use any CSV that helps you detect where on the site MDs are missing and where it can be useful to add them); the file provided in the code has been made available on Google Drive (this way we can always look at the data before running the script).<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><span style=\"font-weight: 400;\">We analyze the data from the crawler and build a dataframe using Pandas.<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><span style=\"font-weight: 400;\">We then choose what URLs are more critical: in the code provided I basically work on the analysis of the wordlift.io website and focus only on content from the English blog that has already a ranking position. Feel free <i><span style=\"font-weight: 400;\">to play<\/i><span style=\"font-weight: 400;\"><\/span><\/span><\/span> with the Pandas filters and to infuse your own SEO knowledge and experience to the script.<\/li>\r\n \t<li style=\"font-weight: 400;\"><span style=\"font-weight: 400;\">We then crawl each page (and here you might want to define the CSS class that the site uses in the HTML to detect the body of the article &#8211; hence preventing you from analyzing menus and other unnecessary elements in the page).<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><span style=\"font-weight: 400;\">We ask BERT (with a vanilla configuration that you can fine-tune) to generate a summary for each page and to write it on a csv file.<\/span><\/li>\r\n \t<li style=\"font-weight: 400;\"><span style=\"font-weight: 400;\">With the resulting CSV we can head back to our beloved CMS and find the best way to import the data (you might want to curate BERT\u2019s suggestions before actually going live with it &#8211; once again &#8211; most of the cases we can do better then the machine).<\/span><\/li>\r\n<\/ol>\r\n<span style=\"font-weight: 400;\">Super easy, not too intensive in computational terms and&#8230;environmentally friendly ?<\/span>\r\n\r\n<span style=\"font-weight: 400;\">Have fun playing with it! Always remember, <\/span><b>it is a robot friend and not a real replacement of your precious work<\/b><span style=\"font-weight: 400;\">. BERT can do the heavy lifting of reading the page and highlighting what matters the most but it might still fail in getting the right length or in adding the proper CTA (i.e. \u201cread more to find \u2026\u201d).<\/span>\r\n<h2><span style=\"font-weight: 400;\">Final thoughts and future work<\/span><\/h2>\r\n<span style=\"font-weight: 400;\">The beauty of automation and agentive SEO is in general, as I like to call it, that you gain super powers while still remaining in full control of the process. AI is far from being magic or becoming (at least in this context) a replacement for content writers and SEOs, rather AI is a smart assistant that can augment our work.\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">There are some clear limitations with extractive text summarization that are related to the fact that we deal with sentences and if we have long sentences in our web page, we will end up having a snippet that is far too long to become a perfect meta description. I plan to keep on working to fine-tune the parameters to get the best possible results in terms of expressiveness and length but&#8230;so far <\/span><b>only a 10-15% is good enough<\/b><span style=\"font-weight: 400;\"> and doesn\u2019t require any extra update from our <\/span><i><span style=\"font-weight: 400;\">natural intelligence<\/span><\/i><span style=\"font-weight: 400;\">. A vast majority of the summaries look good and it is substantial but still goes beyond the 160 character limits.\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">There is, of course, a lot of potential in these summaries beyond the generation of meta descriptions for SEO\u00a0 &#8211; we can for instance create a \u201cfeatured snippet\u201d type of experience to provide relevant abstracts to the readers. Moreover, if the tone of the article is conversational enough, the summary might also become a <\/span><i><span style=\"font-weight: 400;\">speakable paragraph<\/span><\/i><span style=\"font-weight: 400;\"> that we can use to introduce the content on voice-enabled devices (i.e. \u201cwhat is the latest WordLift article about?\u201d). So, while we can\u2019t let the machine really run the show alone, there is a concrete value in using BERT for summarization.\u00a0<\/span>\r\n<h3><span style=\"font-weight: 400;\">Credits<\/span><\/h3>\r\n<span style=\"font-weight: 400;\">As you arrived to the end of this long article, it is time to remind us all that none of this could be possible without the work of many people and enlightened organizations that are committed to open source technologies and that are enabling and encouraging practitioners around the world to make (well, hopefully) the web a better place!\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">It is also thanks to mavericks and SEOs with a data-driven mindset like <\/span><a href=\"https:\/\/twitter.com\/fighto\"><span style=\"font-weight: 400;\">Paul Shapiro<\/span><\/a><span style=\"font-weight: 400;\"> and <\/span><a href=\"https:\/\/twitter.com\/hamletbatista\"><span style=\"font-weight: 400;\">Hamlet<\/span><\/a><span style=\"font-weight: 400;\"> that I got interested in the topic and ready to experiment with new tools!\u00a0<\/span>\r\n\r\n<span style=\"font-weight: 400;\">Give a spin to <\/span><a href=\"https:\/\/colab.research.google.com\/drive\/1TeASgjfCGiZY7VjxcQmHIHHjy5L9Z-j1\"><span style=\"font-weight: 400;\">the code on the Google Colab<\/span><\/a><span style=\"font-weight: 400;\"> and send me any comments or suggestions over Twitter or LinkedIn!<\/span>\r\n\r\n<span style=\"font-weight: 400;\">Want to scale your marketing efforts with Woorank and WordLift <\/span><a href=\"https:\/\/wordlift.io\/seo-management-service\/\"><span style=\"font-weight: 400;\">SEO management service<\/span><\/a><span style=\"font-weight: 400;\">? I can\u2019t wait to learn more about your challenges!\u00a0<\/span>\r\n\t\t\r\n\t\r\n\r\n\r\n\r\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":6,"featured_media":18403,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"wl_entities_gutenberg":"","_wlpage_enable":"yes","footnotes":""},"categories":[612,8],"tags":[],"wl_entity_type":[30],"coauthors":[],"class_list":["post-11859","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-semantic-seo","category-seo","wl_entity_type-article"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v22.6 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>How to write meta descriptions using BERT - WordLift Blog<\/title>\n<meta name=\"description\" content=\"Writing meta descriptions for several webpages has now become much easier thanks to Google&#039;s new natural language processor, BERT.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert\" \/>\n<link rel=\"next\" href=\"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/2\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"How to write meta descriptions using BERT\" \/>\n<meta property=\"og:description\" content=\"Writing meta descriptions for several webpages has now become much easier thanks to Google&#039;s new natural language processor, BERT.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert\" \/>\n<meta property=\"og:site_name\" content=\"WordLift Blog\" \/>\n<meta property=\"article:published_time\" content=\"2019-11-29T10:04:16+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2021-11-03T13:54:18+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert-reading.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1000\" \/>\n\t<meta property=\"og:image:height\" content=\"600\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Andrea Volpini\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Andrea Volpini\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"12 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#article\",\"isPartOf\":{\"@id\":\"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/\"},\"author\":{\"name\":\"Andrea Volpini\",\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#\/schema\/person\/574352082cc71dab8d164410f1cabe0a\"},\"headline\":\"How to write meta descriptions using BERT\",\"datePublished\":\"2019-11-29T10:04:16+00:00\",\"dateModified\":\"2021-11-03T13:54:18+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/\"},\"wordCount\":2274,\"publisher\":{\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#organization\"},\"image\":{\"@id\":\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage\"},\"thumbnailUrl\":\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg\",\"articleSection\":[\"Semantic SEO\",\"seo\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/\",\"url\":\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert\",\"name\":\"How to write meta descriptions using BERT - WordLift Blog\",\"isPartOf\":{\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage\"},\"image\":{\"@id\":\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage\"},\"thumbnailUrl\":\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg\",\"datePublished\":\"2019-11-29T10:04:16+00:00\",\"dateModified\":\"2021-11-03T13:54:18+00:00\",\"description\":\"Writing meta descriptions for several webpages has now become much easier thanks to Google's new natural language processor, BERT.\",\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage\",\"url\":\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg\",\"contentUrl\":\"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg\",\"width\":1200,\"height\":675},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#website\",\"url\":\"https:\/\/wordlift.io\/blog\/en\/\",\"name\":\"WordLift Blog\",\"description\":\"AI-Powered SEO\",\"publisher\":{\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/wordlift.io\/blog\/en\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#organization\",\"name\":\"WordLift\",\"url\":\"https:\/\/wordlift.io\/blog\/en\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/mk0wordliftblog7j5te.kinstacdn.com\/wp-content\/uploads\/sites\/3\/2017\/04\/logo-1.png\",\"contentUrl\":\"https:\/\/mk0wordliftblog7j5te.kinstacdn.com\/wp-content\/uploads\/sites\/3\/2017\/04\/logo-1.png\",\"width\":152,\"height\":40,\"caption\":\"WordLift\"},\"image\":{\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#\/schema\/logo\/image\/\"}},{\"@type\":\"Person\",\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#\/schema\/person\/574352082cc71dab8d164410f1cabe0a\",\"name\":\"Andrea Volpini\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/wordlift.io\/blog\/en\/#\/schema\/person\/image\/466a1652833e48ca11c81b363eba7c25\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/6b9d3d311b50a8749201fe4b318907a8?s=96&d=mm&r=pg\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/6b9d3d311b50a8749201fe4b318907a8?s=96&d=mm&r=pg\",\"caption\":\"Andrea Volpini\"}}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"How to write meta descriptions using BERT - WordLift Blog","description":"Writing meta descriptions for several webpages has now become much easier thanks to Google's new natural language processor, BERT.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert","next":"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/2\/","og_locale":"en_US","og_type":"article","og_title":"How to write meta descriptions using BERT","og_description":"Writing meta descriptions for several webpages has now become much easier thanks to Google's new natural language processor, BERT.","og_url":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert","og_site_name":"WordLift Blog","article_published_time":"2019-11-29T10:04:16+00:00","article_modified_time":"2021-11-03T13:54:18+00:00","og_image":[{"width":1000,"height":600,"url":"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/bert-reading.jpg","type":"image\/jpeg"}],"author":"Andrea Volpini","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Andrea Volpini","Est. reading time":"12 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#article","isPartOf":{"@id":"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/"},"author":{"name":"Andrea Volpini","@id":"https:\/\/wordlift.io\/blog\/en\/#\/schema\/person\/574352082cc71dab8d164410f1cabe0a"},"headline":"How to write meta descriptions using BERT","datePublished":"2019-11-29T10:04:16+00:00","dateModified":"2021-11-03T13:54:18+00:00","mainEntityOfPage":{"@id":"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/"},"wordCount":2274,"publisher":{"@id":"https:\/\/wordlift.io\/blog\/en\/#organization"},"image":{"@id":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage"},"thumbnailUrl":"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg","articleSection":["Semantic SEO","seo"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/wordlift.io\/blog\/en\/write-meta-descriptions-bert\/","url":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert","name":"How to write meta descriptions using BERT - WordLift Blog","isPartOf":{"@id":"https:\/\/wordlift.io\/blog\/en\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage"},"image":{"@id":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage"},"thumbnailUrl":"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg","datePublished":"2019-11-29T10:04:16+00:00","dateModified":"2021-11-03T13:54:18+00:00","description":"Writing meta descriptions for several webpages has now become much easier thanks to Google's new natural language processor, BERT.","inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.woorank.com\/en\/blog\/how-to-write-meta-descriptions-using-bert#primaryimage","url":"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg","contentUrl":"https:\/\/wordlift.io\/blog\/en\/wp-content\/uploads\/sites\/3\/2019\/11\/24-1.jpg","width":1200,"height":675},{"@type":"WebSite","@id":"https:\/\/wordlift.io\/blog\/en\/#website","url":"https:\/\/wordlift.io\/blog\/en\/","name":"WordLift Blog","description":"AI-Powered SEO","publisher":{"@id":"https:\/\/wordlift.io\/blog\/en\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/wordlift.io\/blog\/en\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/wordlift.io\/blog\/en\/#organization","name":"WordLift","url":"https:\/\/wordlift.io\/blog\/en\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/wordlift.io\/blog\/en\/#\/schema\/logo\/image\/","url":"https:\/\/mk0wordliftblog7j5te.kinstacdn.com\/wp-content\/uploads\/sites\/3\/2017\/04\/logo-1.png","contentUrl":"https:\/\/mk0wordliftblog7j5te.kinstacdn.com\/wp-content\/uploads\/sites\/3\/2017\/04\/logo-1.png","width":152,"height":40,"caption":"WordLift"},"image":{"@id":"https:\/\/wordlift.io\/blog\/en\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/wordlift.io\/blog\/en\/#\/schema\/person\/574352082cc71dab8d164410f1cabe0a","name":"Andrea Volpini","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/wordlift.io\/blog\/en\/#\/schema\/person\/image\/466a1652833e48ca11c81b363eba7c25","url":"https:\/\/secure.gravatar.com\/avatar\/6b9d3d311b50a8749201fe4b318907a8?s=96&d=mm&r=pg","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/6b9d3d311b50a8749201fe4b318907a8?s=96&d=mm&r=pg","caption":"Andrea Volpini"}}]}},"_wl_alt_label":[],"wl:entity_url":"http:\/\/data.wordlift.io\/wl0216\/post\/how_to_write_meta_descriptions_using_bert","_links":{"self":[{"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/posts\/11859"}],"collection":[{"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/users\/6"}],"replies":[{"embeddable":true,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/comments?post=11859"}],"version-history":[{"count":0,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/posts\/11859\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/media\/18403"}],"wp:attachment":[{"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/media?parent=11859"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/categories?post=11859"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/tags?post=11859"},{"taxonomy":"wl_entity_type","embeddable":true,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/wl_entity_type?post=11859"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/wordlift.io\/blog\/en\/wp-json\/wp\/v2\/coauthors?post=11859"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}