Frequently asked questions

SemanticIS is a semantic search engine for Information Systems (IS) research articles. Unlike other search engines, SemanticIS provides access to the semantic metadata of research articles, such as domain-specific topics, methods, and theories. Based on this metadata, SemanticIS provides three unique features:

1. Semantic Filters

Users can choose from a variety of semantic filters and filter the available body of knowledge for specific aspects, e.g., all articles that conducted case studies, used a specific theory, or studied an IS-specific topic.

2. Semantic Search Results

Users can more quickly assess whether an article is relevant or not. SemanticIS displays the top 5 most common topics, theories, and methods of each article in the search results. By clicking on any of these tags, SemanticIS displays the exact sentences in which related keywords appear. In addition, SemanticIS offers a tab for "Analytical Results". This tab shows topics, methods, and theories and the number of articles that contain specific key terms. This analysis can be used to inform a drill-down by using additional semantic filters. For example, if a user is interested in the topic "blockchain", the respective semantic filter can be set. All visuals in the analytical results-tab will adjust so that they only show key terms that are present in the filtered articles. The bar graphs can be used to spot additional filters of interest. Each of the key terms from the bar graphs can be used as an additional semantic filter. Thereby, it becomes easy to create a filter funnel.

3. Semantic Analysis

Users can create collections of documents by using the "Save" button below each search result. SemanticIS automatically groups the selected articles based on semantic metadata. This makes it easier to see what methods, theories, and topics are included in a collection of documents.

Semantic filters can be used to search for domain-specific terminology in the full text of research articles. SemanticIS provides an autocomplete field on the search page to find relevant filters, and alternatively, a browser for semantic filters. Users can set multiple filters simultaneously. SemanticIS then returns articles that contain that specific combination of filters. Semantic filters are implemented based on two aspects: the detection of domain-specific keywords and the classification of sentences.

Detection of Domain-Specific Keywords

To identify domain-specific terminology that can be used for semantic analysis, we performed a keyword search of all articles in our database. We searched for all keywords contained in a domain-specific ontology for information systems (anonymized for review). The implemented ontology is a keyword repository that contains approximately 3,000 key terms and 300,000 synonyms for IS-specific methods, theories, or topics.

Sentence Classification

In addition, we have performed sentence classification. By classifying the sentences of an article, we only find key terms that can be attributed to a focal article. Consider the sentences: "We conducted a case study," "Smith et al. (2022) conducted a case study," and "A case study is a research method." Similar to the first example sentence, we only consider sentences that clearly describe a focal article. Therefore, we implemented a transformer-based sentence classification procedure (anonymized for review).

Browser for Semantic Filters

We have implemented a browser for semantic filters that provides an overview of all semantic filters in our database. Any term can be used as a semantic filter by clicking on it. For example, filtering for "case study" will return all articles that contain the term "case study" or any of its synonyms. Clicking on the i icon for any term will bring up an overview of all synonyms.

The following table provides an overview of options for the search syntax.

SemanticIS indexes articles from major IS journals and conference. The following visualizations provide the count of articles by year and by outlet.

Overview of acronyms for outlets.

Each article can be saved into a collection of articles. To do this, users can click the "Save" button under each title in the search results. This will result in the display of an article count button in the upper right corner of the screen. Clicking on this button will take the user to their collection of saved articles.

In the article overview, SemanticIS displays a bar chart showing the distribution of detected topics in the saved articles. The bar graph can display various semantic metadata, such as topics, methods, and theories. Clicking on any bar of the bar graph filters the result table below. This allows users to select articles that share the same semantic metadata.

Below the bar graph, article details are displayed in a results table. Unlike the search results page, where SemanticIS displays only the five most common keywords across all metadata categories, this table displays all extracted metadata to help users decide whether to include or exclude an article.

Citation counts are based on our database. We count how often an article is referenced by other articles in our database. These counts might differ from citation counts in other search engines.