{"id":53894,"date":"2022-01-19T16:15:24","date_gmt":"2022-01-19T16:15:24","guid":{"rendered":"https:\/\/sii.pl\/?post_type=case-study&#038;p=53894"},"modified":"2025-04-08T10:10:50","modified_gmt":"2025-04-08T10:10:50","slug":"ai-based-information-extraction-and-analytics-for-the-construction-market-using-nlp-ml","status":"publish","type":"case-study","link":"https:\/\/sii.pl\/en\/case-study\/ai-based-information-extraction-and-analytics-for-the-construction-market-using-nlp-ml\/","title":{"rendered":"Large volume document management for the construction industry"},"content":{"rendered":"<h2>The challenge<\/h2>\n<div class=\"sii-rl-content-item-value sii-rl-businessNeed-value sii-rl-content-item-value-display\">\n<ul>\n<li><span class=\"TextRun SCXW140960786 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW140960786 BCX8\">The customer processes around 500,000 multipage documents <\/span><span class=\"NormalTextRun SCXW140960786 BCX8\">monthly<\/span><span class=\"NormalTextRun SCXW140960786 BCX8\">, consisting of various layouts and formats. <\/span><\/span><\/li>\n<li><span class=\"TextRun SCXW140960786 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW140960786 BCX8\">Manual processing is slow, costly, and prone to human errors. <\/span><\/span><\/li>\n<li><span class=\"TextRun SCXW140960786 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW140960786 BCX8\">Automating document comprehension and calassification can increase operational efficiency, while a<span class=\"TextRun SCXW230563364 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW230563364 BCX8\">\u00a0knowledge management system provides easy access to key insights.<\/span><\/span><span class=\"EOP SCXW230563364 BCX8\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335551550&quot;:6,&quot;335551620&quot;:6,&quot;335559738&quot;:240,&quot;335559739&quot;:240,&quot;335559740&quot;:360}\">\u00a0<\/span><\/span><\/span><\/li>\n<\/ul>\n<\/div>\n<h2>What we did<\/h2>\n<ul>\n<li><span class=\"TextRun SCXW137317573 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW137317573 BCX8\">We implemented an AI-powered solution using Language Independent Layout Transformers<\/span> <span class=\"NormalTextRun CommentHighlightPipeRestRefresh SCXW137317573 BCX8\">. Our system enables automated extraction of data from both structured and unstructured documents to improve workflows<\/span><span class=\"NormalTextRun SCXW137317573 BCX8\"> reaching over 90% accuracy<\/span><span class=\"NormalTextRun SCXW137317573 BCX8\">.<\/span><\/span><span class=\"EOP SCXW137317573 BCX8\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335551550&quot;:6,&quot;335551620&quot;:6,&quot;335559738&quot;:240,&quot;335559739&quot;:240,&quot;335559740&quot;:360}\">\u00a0<\/span><\/li>\n<li><span class=\"TextRun SCXW229226744 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW229226744 BCX8\">We applied machine learning <\/span><span class=\"NormalTextRun SCXW229226744 BCX8\">and natural language processing <\/span><span class=\"NormalTextRun SCXW229226744 BCX8\">algorithms to automate the document classification process. By training models on historical data, the system categorizes incoming documents with<\/span><span class=\"NormalTextRun SCXW229226744 BCX8\"> 98% <\/span><span class=\"NormalTextRun SCXW229226744 BCX8\">accuracy and minimizes the need for human intervention.<\/span><\/span><span class=\"EOP SCXW229226744 BCX8\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335551550&quot;:6,&quot;335551620&quot;:6,&quot;335559738&quot;:240,&quot;335559739&quot;:240,&quot;335559740&quot;:360}\">\u00a0<\/span><\/li>\n<li><span class=\"TextRun SCXW14213793 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW14213793 BCX8\">We implemented a solution that extracts key information from documents and stores it in a structured, searchable knowledge base. Using contrastive learning techniques and deep neural networks, the system <\/span><span class=\"NormalTextRun SCXW14213793 BCX8\">is able to<\/span><span class=\"NormalTextRun SCXW14213793 BCX8\"> find and retrieve information with high accuracy<\/span><span class=\"NormalTextRun SCXW14213793 BCX8\">, but also to manage the knowledge by enriching existing records with <\/span><span class=\"NormalTextRun SCXW14213793 BCX8\">new information<\/span><span class=\"NormalTextRun SCXW14213793 BCX8\"> when new evidence appea<\/span><span class=\"NormalTextRun SCXW14213793 BCX8\">r<\/span><span class=\"NormalTextRun SCXW14213793 BCX8\">s<\/span><span class=\"NormalTextRun SCXW14213793 BCX8\">.<\/span><\/span><span class=\"EOP SCXW14213793 BCX8\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335551550&quot;:6,&quot;335551620&quot;:6,&quot;335559738&quot;:240,&quot;335559739&quot;:240,&quot;335559740&quot;:360}\">\u00a0<\/span><\/li>\n<li><span class=\"TextRun BCX8 SCXP135040644\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun BCX8 SCXP135040644\">We implemented an innovative AI agent system that proactively enhances data <\/span><\/span><span class=\"TextRun BCX8 SCXP135040644\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun BCX8 SCXP135040644\">quality by automatically searching online sources to find and integrate missing or <\/span><\/span><span class=\"TextRun BCX8 SCXP135040644\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun BCX8 SCXP135040644\">incomplete information.<\/span><\/span><\/li>\n<\/ul>\n<h2>Benefits for the client<\/h2>\n<ul>\n<li><span class=\"TextRun SCXW31880414 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW31880414 BCX8\">The client experienced better data accuracy, reduced manual effort, and increased operational speed \u2013 which allowed them to scale their document processing capabilities<\/span><\/span><span class=\"TrackChangeTextInsertion TrackedChange SCXW31880414 BCX8\"><span class=\"TextRun SCXW31880414 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW31880414 BCX8\"> by over 80<\/span><\/span><\/span><span class=\"TrackChangeTextInsertion TrackedChange SCXW31880414 BCX8\"><span class=\"TextRun SCXW31880414 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW31880414 BCX8\">%<\/span><\/span><\/span><span class=\"TextRun SCXW31880414 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW31880414 BCX8\">.<\/span><\/span><span class=\"EOP SCXW31880414 BCX8\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335551550&quot;:6,&quot;335551620&quot;:6,&quot;335559738&quot;:240,&quot;335559739&quot;:240,&quot;335559740&quot;:360}\">\u00a0<\/span><\/li>\n<li><span class=\"TextRun SCXW18027560 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW18027560 BCX8\">The client achieved fast<\/span><span class=\"NormalTextRun SCXW18027560 BCX8\"> and <\/span><span class=\"NormalTextRun SCXW18027560 BCX8\">accurate<\/span> <span class=\"NormalTextRun SCXW18027560 BCX8\">document processing <\/span><span class=\"NormalTextRun SCXW18027560 BCX8\">solution <\/span><span class=\"NormalTextRun SCXW18027560 BCX8\">outperforming humans (98% accuracy)<\/span><span class=\"NormalTextRun SCXW18027560 BCX8\"> \u2013 which led to cost savings and improved operational efficiency.<\/span><\/span><span class=\"EOP SCXW18027560 BCX8\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335551550&quot;:6,&quot;335551620&quot;:6,&quot;335559738&quot;:240,&quot;335559739&quot;:240,&quot;335559740&quot;:360}\">\u00a0<\/span><\/li>\n<li><span class=\"TextRun SCXW188404968 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW188404968 BCX8\">The <span class=\"TextRun BCX8 SCXP117225890\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun BCX8 SCXP117225890\">solution revolutionized knowledge management outperforming human <\/span><\/span><span class=\"TextRun BCX8 SCXP117225890\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun BCX8 SCXP117225890\">experts reaching over 92% accuracy, enabling our customer to develop the <\/span><\/span><span class=\"TextRun BCX8 SCXP117225890\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun BCX8 SCXP117225890\">market&#8217;s most comprehensive and current construction projects database.<\/span><\/span><\/span><\/span><\/li>\n<li><span class=\"TextRun SCXP267431941 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun SCXP267431941 BCX8\">Increased completeness of the information in customer\u2019s CRM system by <\/span><\/span><span class=\"TextRun SCXP267431941 BCX8\" lang=\"EN-US\" xml:lang=\"EN-US\" data-scheme-color=\"@595959,1,18:65000,19:35000\" data-usefontface=\"true\" data-contrast=\"none\"><span class=\"NormalTextRun SCXP267431941 BCX8\">40% letting for more effective business processes.<\/span><\/span><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>The challenge The customer processes around 500,000 multipage documents monthly, consisting of various layouts and formats. Manual processing is slow, &hellip; <a class=\"continued-btn\" href=\"https:\/\/sii.pl\/en\/case-study\/ai-based-information-extraction-and-analytics-for-the-construction-market-using-nlp-ml\/\">Continued<\/a><\/p>\n","protected":false},"author":39,"featured_media":117299,"template":"views\/single-old-case-study.blade.php","offering":[5112,4152,1474],"industry":[1660],"client":[5098],"technologies":[2368,5100,5109,5103,5106,1806],"country":[],"class_list":["post-53894","case-study","type-case-study","status-publish","has-post-thumbnail","hentry","offering-artificial-intelligence","offering-data-analytics","offering-digital","industry-retail-e-commerce","client-dodge-en","technologies-aws","technologies-deep-learning-en","technologies-flask-en","technologies-machine-learning-en","technologies-natural-language-processing-en","technologies-python"],"acf":[],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/case-study\/53894"}],"collection":[{"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/case-study"}],"about":[{"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/types\/case-study"}],"author":[{"embeddable":true,"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/users\/39"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/media\/117299"}],"wp:attachment":[{"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/media?parent=53894"}],"wp:term":[{"taxonomy":"offering","embeddable":true,"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/offering?post=53894"},{"taxonomy":"industry","embeddable":true,"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/industry?post=53894"},{"taxonomy":"client","embeddable":true,"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/client?post=53894"},{"taxonomy":"technologies","embeddable":true,"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/technologies?post=53894"},{"taxonomy":"country","embeddable":true,"href":"https:\/\/sii.pl\/en\/wp-json\/wp\/v2\/country?post=53894"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}