Academic Writing AdviceAcademic, Writing, Advice
ServiceScape Incorporated
ServiceScape Incorporated
2020

Microsoft Academic Search (MAS) Review: Everything You Need to Know

ScienceEditor

Microsoft Academic Search is a freely available website that helps scholars find the most relevant scholarly information. Microsoft Academic Search (MAS) is designed as a semantic search engine, which means that it aims to find the best matches for the user's meaning or intent, and does not just match keywords. While other search engines have also moved in this direction, Microsoft Academic further applies machine learning to better characterize and sort web content to provide the most useful results.

Why is searching for scholarly information so challenging?

Let's consider a misspelled Google search for "heit shock protein", which is automatically redirected to a search for "heat shock protein". The top results include basic information about heat shock protein, and "Scholarly articles for heat shock protein", which links directly to the results in Google Scholar. Google Scholar is another freely available website for scholarly information, and as of October 2020 gives approximately 2,370,000 results for the term "heat shock protein." Next to key information about each article (title, authors, journal, and first three lines of the abstract), the search results may include a link to an open access version of the full text article, which can be very useful.

Google Scholar presents the most prominent results first, but each article for "heat shock protein" provides just a snapshot of all of the research being done for this wide-ranging topic. The search results do not make it easy to identify related concepts, to quickly understand the major advances that have occurred in this research area, or to identify the most prominent scholars in the field. Using PubMed to search for "heat shock protein" results in a list that is smaller (74,213 articles as of October 2020) and has more filtering options, but is still overwhelming.

What is unique about Microsoft Academic Search (MAS)?

Microsoft Academic aims to organize and sort scholarly information, so that the most relevant results are presented first, and related concepts can be easily accessed. A search for "heat shock protein" on Microsoft Academic leads to more than 50,000 results (as of October 2020). In the left sidebar, there are multiple filtering options that are specific to the current search. For "heat shock protein", the "Top Topics" in the left sidebar include Cell Biology, Chemistry, Medicine, and the protein Hsp70. Clicking the "More" button reveals additional subcategories including Genetics, and Cancer research.

This machine-curated list of subtopics in the left sidebar is what really differentiates Microsoft Academic Search from Google Scholar and PubMed. You can start with a general idea of what you are interested in, and MAS will suggest possible subcategories that enable you to quickly find the papers that are most interesting to you. In Google Scholar and PubMed, you need to add additional search terms to focus your search. This is challenging because you often don't know what the most effective words will be, and because the relevant terms might not be included in the meta-data for a relevant journal article or website. For example, a paper published in a chemistry journal might not list "chemistry" as a key word. Microsoft Academic aids to eliminate problems due to idiosyncrasies in word choice through machine learning.

Six Main Categories Used by Microsoft Academic
The icons used to identify the six main categories used by Microsoft Academic to organize data.

How can I narrow my search in MAS?

In Microsoft Academic Search, the left sidebar also allows you to filter by "Publication Types", which can include Journal publications, Patents, Books, Book chapters, and Conference publications. These publication types are also covered by Google Scholar, while PubMed does not include patents or informal publications like conference publications. The coverage of books, book chapters, working papers, and technical reports can be especially important in the social sciences. The left sidebar also allows you to filter by "Top Authors", "Top Journals", "Top Institutions", and 'Top Conferences", and "Time" (e.g. year of publication). These search filters can be clicked in numerous combinations to narrow your search. After a filter is clicked, other filters may disappear (e.g. because they would no longer produce any results) and new ones may take their place. These dynamic updates provide information about the options available within the current search.

Narrowing the search can also lead to additional information being shown on Microsoft Academic Search. For example, from the original search for "heat shock protein", I can use the left sidebar to choose "Hsp70" under "Top Topics". This causes a new box to appear in the right sidebar that includes a picture and brief description of the protein Hsp70. The box also shows "Parent Topics" (more general topics), "Child Topics" (more specific topics), and "Related topics".

The Right Sidebar and the Center Column
The right sidebar includes boxes for the main topic and related topics. The center column includes publications currently sorted by relevance.

What are the benefits of machine learning in MAS?

Clicking on any topic marked with a laboratory flask icon—such as "Hsp70" in the right sidebar—leads to a page for that topic. These pages are produced by Microsoft Academic algorithms, and contain some very useful information. The top of the page for "Hsp70" provides basic information, and things quickly become more detailed as you scroll down. There is a graph of "Publications & Citations Over Time" for this topic, followed immediately by tabs for "Publications", "Authors", "Conferences", "Journals", and "Institutions." Clicking on the tab for "Authors" will reveal a graph of the most impactful authors for this topic, with larger circles representing more impactful authors. The X axis represents Rank by saliency (i.e. impact), while the Y axis represents Rank by saliency per family (i.e. impact for this particular research topic).

Graph
Clicking on the tabs for "Conferences", "Journals", and "Institutions" reveal similar graphs for these categories. You can click on any of the circles within a graph to show relevant publications in Microsoft Academic.

Let's return to the "Publications" tab on the page for Hsp70, which has 15,594 results (as of October 2020). In the left sidebar, you can still filter by Time (e.g. year of publication), Top Topics, Top Authors, Top Journals, Top Institutions, and more. In the upper right of the center column, you can see that the publication list is initially sorted by "Relevance". You also also sort by "Saliency" (i.e. Impact), "Newest first", "Oldest first", and "Most estimated citations". Since a publication needs time to accumulate citations in other publications, "Saliency" is believed to be the best measure of a publication's impact. At the bottom of the box for each publication, you can click on the icon that looks like a stack of three books to add the publication to your reading list.

For each entry in the "Publications" list, many of the text phrases are links. Clicking on the name of an institution (e.g "University of Groningen") takes you to the Microsoft Academic page about that institution, with associated Publications, Authors, Conferences, etc. The same is true of each author name, journal title, and topic.

What information does MAS provide about individual publications?

Clicking on a Publication title (e.g. "HSP70 chaperone machinery: J proteins as drivers of functional specificity") brings you to the Microsoft Academic page for that article, which includes the full abstract. The page also includes the references included within the paper, other publications that have cited the paper, and related publications, all of which can be filtered by the tools in the left sidebar.

While many aspects of Microsoft Academic are quite nice, the process of obtaining a full text version of an article can be quite clunky. Once you have identified a publication of interest within a list, you must click on the title to open the Microsoft Academic page for that publication. Below the title, there is usually a DOI number (Digital Object Identifier), which links directly to the article on the publisher's website. For many journal articles, there is a paywall that prevents access to the full text version without an institutional affiliation, personal subscription, or other payment.

Going back to the Microsoft Academic page for the article of interest, there are usually "Other Links" listed under the Abstract. Somethes, there is a link for "View PDF", which takes you to an open access version of the full text pdf. More often, you would need to explore the links for "Website(s)" or "Additional link(s)" to try and find a full text version. If this is the case, I recommend searching for the publication directly in Google Scholar, PubMed, or another platform that provides easier access to full text publications.

In summary, Microsoft Academic Search is a freely available website that can greatly simplify the process of identifying the most relevant scholarly information for a topic of interest, but is not the best source for finding links to full text. As the amount of scholarly information continues to grow at an ever-increasing pace, machine learning will play an ever-increasing role in identifying relevant information.

Get in-depth guidance delivered right to your inbox.
Subscribe