Indexing Interviews in OHMS

Indexing Interviews in OHMS: An Overview

By Doug Boyd, Danielle Gabbard, Sarah Price, and Alana Boltz

 Overview

OHMS is an online tool for enhancing access to online oral histories created by the Louie B. Nunn Center for Oral History at the University of Kentucky Libraries.  Through a grant form the IMLS (Institute of Museum and Library Services), the Nunn Center has made OHMS open source and freely available.  Discovery of, and access to oral history in an archival setting has been challenging.  Oral history involves audio or video recording on time-based media.  Unlike print or photographic resources, it is cumbersome for users or researchers to visually scan an oral history for content unless a transcript is created, textually representing the recorded event.  In the absence of a verbatim transcript, users must listen to each moment in an interview in order to determine the potential for relevant information discovery.  Until speech recognition and artificial intelligence technologies become more accessible and affordable, transcription remains difficult and expensive.

When conducting fieldwork or interviews in an analog recording context (audio cassette and reel to reel tape), oral historians, folklorists, anthropologists and other related practitioners would sequentially “log” their tapes on paper to create efficient ways to get back to the moments that they determined were “important” or “useful.”  These logs would include  a variety of metadata types, including partial transcripts, titles, keywords and descriptions, which all corresponded to a counter number. Analog technologies lacked true time code and were measured by a counter that was usually machine specific.  While these tape logs were essential for the researcher who was writing a book or an article, this usefulness rarely transferred into an archival context.  As formats were migrated, the counter numbers (which were specific to each machine) failed to maintain consistency.  As a result, archivists and researchers utilized the tape logs instead as the source for comprehensive item-level metadata.  However, the archivist/user/researcher still had to manually scan the audio/video in order to access the corresponding segments represented in the log.

OHMS creates a method to index, log, or annotate an oral history interview with segment-level descriptive metadata that corresponds to true timecode in a way that translates to the user side of the oral history process.  In the article OHMS: Enhancing Access to Oral History for Free, OHMS creator and Nunn Center Director Doug Boyd poses a useful analogy pertaining to OHMS indexing and the user experience:

If you were to go to the grocery store and ask the manager, “Where can I find the Cheerios?,” the manager would, most likely, smile and say “aisle ten.” At that point, you would, upon identifying the location of aisle ten, proceed down aisle ten until you were at the specific location of the Cheerios. In many ways, this is how I view the role of indexing an oral history interview.[1]

Another useful analogy for thinking about indexing using OHMS, from a user perspective, is to think about a book.  The OHMS index provides the user/researcher the browse function of a book’s table of contents, while simultaneously providing the granular and precise searchability and the navigational interface of a book’s index.  The result is a user experience that makes an indexed interview browsable and searchable in ways that link the user’s search terms to corresponding moments in the audio or video interview.  For more information about using OHMS go to OHMS website at www.oralhistoryonline.org.

The Nunn Center Guide to Indexing

The primary goal when indexing oral history interviews in OHMS is to create a useful and affordable user interface for oral histories. It is equally important to create a dynamic framework that optimizes the effectiveness and efficiency of information discovery in both single interviews as well as in oral history collections of different sizes.

Indexing is a subjective process and can vary greatly in breadth and depth. The Nunn Center has worked out three levels of indexing, which correspond to the thoroughness of our desired outcome. Level one indexes make use of fewer metadata fields while also requiring less detail for each field. Level one segments include a title and keywords. A level two index provides a good balance between the sparse information of a level one index and the detailed, longer process of creating a level three index. Level two indexes contain a title, keywords and/or subjects, partial transcripts and a brief synopsis. A level three index is the most detailed level, providing users with the greatest amount of information. Level three indexes make use of all available metadata fields.

An index is only as good as the indexer and the effectiveness of the segment or story-level metadata being created.  The following is a guide to assist indexers in maximizing the effectiveness and future usefulness of the oral history indexes they create in OHMS. This is not a rulebook; it is a model containing guidelines we have developed at the Nunn Center over the past year. Our intention is to update this guide in an ongoing fashion.

We have also produced the 25-minute video Using OHMS to Index Oral History: A Detailed Tutorial if you would like to see a video version of this guide with examples.

If you would like to see an example of a strong OHMS index that is referenced throughout this guide, view the interview with Wild Turkey Bourbon Distillery Master Distiller Jimmy Russell from the Nunn Center’s Kentucky Bourbon Tales Oral History Project.  

What gets Indexed?

Each interview contains hundreds of possible access points that could potentially prove useful to researchers someday.  Indexing involves making good decisions about the representation of data in an interview.  Most professional efforts in the archival community focus on creating collection or series level descriptive metadata, and when the resources warrant it, at the item level. OHMS provides a framework for creating access points within the item by constructing searchable, descriptive metadata packages at the segment level that correlate to a corresponding moment in the audio or video recording.

For each segment created in an OHMS index, the indexer can create information in the following fields:

• Time Stamp
• Partial Transcript
• Segment Title (required)
• Subjects (semi-colon delimited)
• Keywords (semi-colon delimited)
• Segment Synopsis
• GPS Coordinates
• GPS Description
• Hyperlink
• Link Description

Oral history interviews are not just random recordings; they are structured interviews that result in a dynamic interaction between interviewer and interviewee. Often, interviews conducted as part of a larger project will contain a natural structure that is driven by emerging patterns created by the interviewer. For example, interviews from the Nunn Center’s From Combat to Kentucky Oral History Project (a collection of interviews with veterans returning from Iraq and Afghanistan who are enrolled in higher education) follow a similar structure from interview to interview:

• Interview introduction
• Biographical information
• Making the decision to join the military
• Enlistment
• Boot camp, training and specialization
• Deployment
• Numerous stories documenting their deployment experiences
• Transition out of the military
• Transition into higher education
• Conclusion

In an archival context, indexing an oral history interview is about effectively documenting and describing interview transitions. Our goal is to provide the future user of the interview with a descriptive search and browse framework. Continuing the book analogy, the indexer creates chapter marks that correlate with these natural transitions, as well as titles and a description for each chapter. Understanding the interview’s structure is crucial for effectively documenting these transitions.

• In an oral history interview, questions often indicate the locations of subject transitions.

• By choosing to index (or not to index) a moment in an oral history interview, the indexer is controlling significance. This is a very powerful position to be in since the index you create will frame all future navigation of that interview. It is useful to ask yourself the following questions when deciding whether to index a moment (or not):

• Does it represent a major topic within this segment of the interview?
• Will inclusion of this topic be useful to future researchers?
• Does this segment contain unique, compelling, or interesting content that, even if discussed briefly, stands out in the interview?
• Does the content correlate to major historical events such as the Great Depression or World War II?
• Will the creation of an index point for this segment represent this interview effectively to a user, researcher or even a search engine?
• Have I created index points that will help a user navigate this interview?
• Have I represented the content effectively?

An OHMS index is not a substitution for a transcript. The OHMS index creates a structured and descriptive navigational framework for effectively and efficiently engaging with an oral history interview in an online archival context. The transcript provides text-based, keyword searchability of an interview’s verbatim content. In an OHMS index, you have the opportunity to map natural language to concepts. For example, an individual could talk about living under segregation for much of a 3-hour interview, yet never mention the word “segregation.” This would require the user of a searchable transcript to make a series of educated guesses about words or phrases that might relate to the topic of segregation in order to determine the interview’s relevance to their information needs. A good index will tell the user that the interview contains a discussion about segregation, even if the word “segregation” is never mentioned in that interview.

How to Index: The Mechanics of Indexing in OHMS

This section has been taken from the OHMS: Getting Started Guide. For more details, please consult that guide.  Activating an interview’s indexing module for the first time will bring up a view such as this one:

In order to begin indexing, you must press the “Play” button on the player. The player must be playing in order for you to create an index point, which is created by pressing the “Tag Now” button at the appropriate moment. When you press “Tag Now” you are presented with a series of empty descriptive fields for you to fill out.

The indexer can control the player within the tagging module. The nature of indexing warrants that the indexer will, inevitably, tag a segment late. In other words, you don’t know something is important until after you have already heard it. For this reason, the player backtracks a few seconds each time the tagging module is activated

When creating an OHMS index, the time stamp is critical. You are creating metadata that corresponds to a particular time code. As a result, you will need to make sure that placement of the time stamp (the precise location in the interview where you want the segment to begin) is exactly where you want it to be. This can be done by controlling the player in the tagging module.

Metadata Fields: Creating an Index

Although we have identified a function for each index field, each institution should discuss specific policies regarding the particular use of these fields. For more information about the Nunn Center’s indexing policies and decision-making process, please view the “Using OHMS to Index Oral History: A Detailed Tutorial” video at OralHistoryOnline.org.

OHMS Metadata fields at the segment level include:

• Time Stamp
• Partial Transcript
• Segment Title (required)
• Subjects (semi-colon delimited)
• Keywords (semi-colon delimited)
• Segment Synopsis
• GPS Coordinates
• GPS Description
• Hyperlink
• Link Description

Time Stamp

This is created as soon as you press the “Tag Now” button, and will determine the location of the starting point for the audio or video segment. However, you will need to adjust your time stamp so that your segment correlates to the appropriate time. Place the time stamp where you think is the most appropriate entry point to a segment from the user’s perspective. Often, a segment beginning corresponds to an interviewer’s question. When indexing in OHMS, the Louie B. Nunn Center for Oral History chooses to place timestamps at the beginning of an interviewer’s question instead of before an interviewee’s response. This allows each segment to be more cleanly broken down by topic and also creates a more complete segment. Be careful that you place the time stamp in a comfortable location, such as at the beginning of a question or answer, rather than mid sentence.

Partial Transcript

The partial transcript field can be used to convey the verbatim transcription of the recorded segment. This can be important because it creates an orientation point for the user of an oral history index. For example, a segment could be titled “Segregation” but the narrator may not address this topic immediately. Even minimal population of this field will transform the user experience. Although some may utilize this tool to fully transcribe the segments, the Nunn Center transcribes only the first 140 or so characters of the segment, following the Nunn Center’s transcribing standards. For example, in a segment titled “Major events in the history of the Wild Turkey Distillery” from an interview with Master Distiller Jimmy Russell:

Example, Full Transcribed Segment:

YOUNG: Well tell me about your distillery. What makes it unique?

RUSSELL: Well, this distillery–you know, we just built a new distillery–the first new distillery, completely, that’s been built in Kentucky for, uh, gosh, I don’t know how many–it’s been a long, long time since comp–every–everybody’s had to remodel and expand because the bourbon business got too good. We were across the road here, and we sit right on top of the Kentucky River and US Highway Sixty-two. And we’d expanded as far as we could expand. So, we had this property up here on the hill, and we come up here and built a complete new distillery. Now, we haven’t changed anything–the way we make it. We’re still using the same formula, the same–all I can vouch for, for our yeast, it’s fifty-nine years old. It was–the yeast–it was here when I got here, and we’re still using the same yeast culture.

Example, Transcript Indicated in the OHMS “Partial Transcript” Field:

Well tell me about your distillery. What makes it unique?

Often the verbal interaction being reflected in the partial transcript will involve multiple speakers.  The Nunn Center has chosen to transcribe only the first speaker’s dialogue, although other repositories may choose different transcript structures that identify speakers when appropriate.  For example:

Young: Well tell me about your distillery. What makes it unique?

Russell: Well, this distillery–you know, we just built a new distillery–

If you have placed the time stamp in a potentially awkward place for partial transcription (several people talking at once), you can adjust the time stamp and make up for the content by utilizing additional fields in the index. For usability purposes, we strive to adjust the time stamp to correspond to a single speaker for this field.

Segment Title

Segment titles are required. It is best to use descriptive titles that act as chapter titles for the interview. The titles in the OHMS Viewer function as a table of contents, offering the user a quick glimpse of the contents of the interview. When creating a title for a segment in OHMS, it is helpful to do so in a way that assumes that if a user/researcher never opens the title tab to explore content further, they would still understand the essence of the content of a segment and of the overall interview. The titles within an OHMS index are the primary access point for browsing the contents of an interview.

• Be as succinct and descriptive as possible.  Use titles that clearly and plainly represent the content of this segment to the future user.
  • Example: Distillery activity is better represented as Wild Turkey’s bourbon making process

• Do not try to be clever or use ambiguous or esoteric references in titles.
  • Example: Good things come in small packages is better represented as The development of the Blanton’s brand and other small-batch bourbons

• Avoid using slang.

• If you struggle with creating titles, ask yourself the following questions:
  • What is the main point of this segment of the interview?
  • How would the topic most effectively be represented to a contemporary audience?
  • Does your title require esoteric information about the topic in order to understand the content it represents?
  • Does the title you chose require clarification?

• The Nunn Center begins indexes with the title Interview introduction, if there is an introductory portion of the interview. Similarly, if there is a concluding portion of the interview, the final segment is titled Interview conclusion.

• The Nunn Center only capitalizes the first word of the title and the proper nouns present in the title.

• You can create title root words such as “Childhood” that represent parts of a longer story or major categories that repeat. They are then qualified by adding a — and a narrower, more specific term or phrase.  For example:
  • Title: Childhood–Growing up in Frankfort, Kentucky
  • Title: Childhood–Family
  • Title: Childhood–Elementary school

• It is unnecessary to use the interviewee’s names in segment titles
  • Title: Russell becoming a Master Distiller should be Becoming a Master Distiller
  • Title: Russell’s personal accomplishments should be Personal accomplishments

The most important question to ask when creating a title is, “What is the main point of the segment?” A good rule of thumb is to consider adding a topic to the title if the interviewee talks about the subject for at least ⅓ of the segment. Although this is not a hard and fast rule, it can be helpful to think about, especially when you first begin learning to index. If the segment does not have one main topic but instead 2 or more smaller topics, consider including more than one topic by separating them with a slash.

• Title: Advice for women in the bourbon industry / cooking with bourbon

 Examples of “Good” Titles

The following are examples of good titles we have seen:

• Title: Biographical background and education
• Title: Roach’s parents (generally interviewee’s names do not need to be included in titles but can be useful for clarification when referring to their family, friends, etc. [i.e., “Roach’s parents” instead of “Parents”])
• Title: Early teaching career
• Title: Race relations in Frankfort
• Title: Becoming the Vice President

 Examples of “Not-So-Good” Titles:

The following are examples of titles that could be improved:

• Title: Narrow escape
• Title: Fisticuffs
• Title: A story about Lincoln and Seward (Should be “A story about President Lincoln and Secretary of State Seward)
• Title: Becoming “the Veep”

Subjects

This field is one of two OHMS indexing fields designed for the entry of descriptive terms that represent the content of each segment. These are crucial for providing effective searchability. The Nunn Center uses this field primarily for Library of Congress Subject Headings, though it may be used for any controlled vocabulary. A lesson on using LCSH is beyond the scope of this guide. However, Library of Congress Subject Headings can be vague and general at the interview level, and the segment level often demands even greater specificity. Nevertheless, we have found them invaluable for interoperability of metadata. The indexer’s goal in creating subjects and keywords is to provide representative and descriptive terms for searchability, but it is also to map natural language to concepts. As mentioned earlier, a narrator/interviewee can talk for 3 hours about living under segregation without ever saying the word “segregation.”

For the Nunn Center, it is important for the indexer to use effective descriptors, such as the Library of Congress Subject Headings: 

Example: African Americans–Segregation–Kentucky

• This field allows for multiple entries, which are separated by a semi-colon. It can be used in conjunction with a thesaurus to control or suggest terms to the indexer (see the OHMS: Getting Started Guide).

• The Nunn Center strictly controls the use of this field. If a term is not an auto-suggested term, it can be either added to the thesaurus before using or added as a keyword.

• If an indexer is not familiar with cataloging standards, we recommend that they use the keywords field exclusively until training has been provided.

 Keywords

Like the subjects field in OHMS, the keywords field contains descriptive terms that represent the content of each segment. These are crucial for searchability. The Nunn Center uses the keywords field for any term that represents or describes the content in the segment. As with the subjects field, the indexer’s goal in creating of keywords is to provide representative and descriptive terms for searchability, and also to map natural language to concepts. Unlike the subject field, keywords are specific and local terms that effectively communicate the content of the segment. Using the segregation example referenced in previous sections, accompanying keywords might include:

Restaurants; Water fountains; Lunch counters; Bathrooms

To determine the keywords that should be assigned to the interview, there are several methods indexers can follow. If a topic makes up at least 20% of a segment, consider adding it as a keyword. Indexers should also add synonyms or variations that you think a user may search for, other terms future researchers may be interested in, and choose parts of the segment that are unique and interesting as keywords.

Users will often search by keywords, so the longer and more complicated our keywords are, the less effective the index will be.

•  Unlike subjects, we use this field liberally. We encourage as much specificity as the indexer deems necessary to effectively communicate the contents of the segment. This should be balanced with brevity in order to maintain searchability.

• When tagging keywords, try to use words that the interviewee uses. If it is a somewhat esoteric term, such as a nickname, a regional term, or an unusual phrase, put it in quotation marks.

• Quotation marks should also be used for terms that are used by the interviewee but are no longer appropriate in a cataloging or metadata context. This is important to retain the original context of an interview and also deploys good metadata practice. For example, interviewees in some of our older interviews, such as those in the Robert Penn Warren Civil Rights Oral History Project, make frequent use of the word “negro”. This is no longer an appropriate term for African Americans in a metadata context. Using quotation marks here indicates that the term is used within the interview, but is not one assigned by the indexer.

• Keywords are an excellent way to balance out the generality of controlled vocabulary by using more specific terms. Believe it or not, until recently there was no Library of Congress Subject Heading for “Bourbon.” The keywords you choose can be an effective way to compensate for an oversight such as this one.

• This field allows for multiple entries, separated by a semi-colon, and can be used in conjunction with a thesaurus to control or suggest terms to the indexer (see the OHMS: Getting Started Guide). The Nunn Center uses this field for local terms and tags as well as project-specific vocabularies.

• The Nunn Center recommends making keywords as short and descriptive as possible, unless, of course, they are proper nouns.

• In general, keywords are only one or two words long. For example, “making bourbon in Kentucky” is descriptive, but so long that it is unlikely to be searched together as a phrase. Instead, break this concept into keywords like “Bourbon”, “Kentucky” and “Distilling”. It is far more likely for a researcher to search one of these terms than that they would use this specific phrase.

• Try to make keywords as specific as you can. For example, the word “fun” might be used by the interviewee, but it is too vague and subjective to be a useful search term. Instead, if possible you should use terms related to the activity they are describing.

• The Nunn Center typically pluralizes generic terms as the Library of Congress does. This enables users searching for either version to locate the term in your interview.

More on Subjects and Keywords

Make sure your subjects and keywords in an OHMS index work together. Both fields are searchable, so there is no reason to represent the same content in both. The descriptive terms you choose should thoroughly represent the content of the segment. As with the interview or collection level, choose words that accurately and comprehensively describe the segment.   When possible, we recommend that you provide indexers with relevant controlled vocabularies (even for keywords) to ensure consistency between interviews and projects.

Segment Synopsis

The segment synopsis is designed to contain a descriptive statement about the segment. Synopses are used to further clarify what the segment is about. The segment synopsis is particularly useful to express concepts or topics that are too complex to be conveyed by keywords alone. A good synopsis complements the other descriptive metadata fields in the OHMS index. For example, if the title of the segment is “Growing up during the Great Depression”, an effective synopsis might “Johnson reminisces about her childhood and the hardships her family experienced. She talks about her parents’ struggle to make ends meet and ways that economic hardship affected her as a child.”

• The Nunn Center typically refers to the interviewee and interviewer by last name in a synopsis. It is too informal to refer to them by their first names, and may be confusing. If both the interviewer and interviewee have the same last name, a first initial should be used to distinguish the two (i.e., A. Kelly and D. Kelly). If the interviewees have both the same last name and first initial, full names should be used (i.e., William Marshall and Willard Marshall).

• Synopses should also be relatively short, typically no longer than a sentence or two. They provide a brief summary, but should not be a substitute for listening to the segment.

• The synopsis field in OHMS is very versatile. It can also be used to provide additional comments or details such as the presence of technical problems in the original recording. These comments should be placed in brackets. Repositories may wish to create a standardized list of these commonly used phrases in order to maintain consistency.

• In general, use synopses to clarify what is stated in the title and describe some of its nuances.

GPS Coordinates

This field connects a user from the oral history segment to a location on Google Maps, though the institution can change the map resource if desired. Coordinates are entered in the format “XX.XXX, YY.YYY” where X is latitude (north or south) and Y is longitude (east or west). Only one set of coordinates per segment is allowed at this time. There must be a space following the comma.

GPS Description

This field serves as a label for the GPS coordinates you have provided. Each institution will need to determine standards for how they want to textually describe the GPS coordinates, but we typically put it in the standard Library of Congress format for locations. The text entered into this field will serve as the linkable text in the OHMS viewer. The image below is how the GPS coordinates and text will appear to users in the OHMS viewer:

Hyperlink

This field allows indexers to include a hyperlink in the segment, connecting the oral history index to an external resource of any type. Links to websites especially should link to a specific page with information related to the segment and not a general homepage or website.

Only one hyperlink is allowed per segment at this time. The full hyperlink must be entered into this field:

Hyperlink Description

This field serves as a label for the hyperlink that you have provided. Each institution will need to determine standards for how they want to textually describe these hyperlinks. As with GPS coordinates, the text entered into this field will serve as the linkable text in the OHMS viewer (as opposed to the raw hyperlink).

Using Controlled Vocabularies and Thesauri

Title, subjects, and keywords fields in OHMS can correlate to an uploaded thesaurus of terms. Each field can be assigned a different thesaurus for a particular interview. If a thesaurus has been uploaded and properly assigned in the metadata manager, terms will be automatically suggested based on a partial keying of letters. If the indexer begins typing “segrega” in the subjects field, all of the terms in the assigned thesaurus containing “segrega” will be suggested. The indexer then selects the term desired, and the term is dropped into the subjects field.

See the thesaurus manager section in the OHMS: Getting Started Guide for more information on uploading and assigning a thesaurus.

 Concluding Thoughts on using OHMS to Index Oral History

An effective segment in an OHMS index should utilize fields so that each metadata type works together to convey an overall meaning. Repetition of terms in these fields can be useful. For example, you could use the term “Segregation” both in a title and as a keyword. However, remember that the fields should also complement each other, with different types of terms in each section. If the term “Segregation” is in the title, then also listed as a subject or a keyword, the term does not necessarily need to be used in other fields as well.

Additionally, the fields in an OHMS index are meant to be flexible. Although the Nunn Center utilizes the subjects field for Library of Congress authorized headings and the keywords field as descriptive tags without an authorized source, we are open to utilizing these fields in alternative ways, depending on the particular project.  As stated above, this is not a rulebook, this is a a set of standards and recommendations we have created at the Nunn Center after several years of using OHMS to index oral history.  This is a living, dynamic document that will grow as OHMS and our indexing workflow matures.

[1] Douglas A. Boyd. “OHMS: Enhancing Access to Oral History for Free” Oral History Review, Volume 40, Issue 1, Summer: Fall 2013.

Citation for Article

APA

Boyd, D. A., & Danielle Gabbard, Sara Price, and Alana Boltz (2014). Indexing Interviews in OHMS: An Overview. In D. Boyd, S. Cohen, B. Rakerd, & D. Rehberger (Eds.), Oral history in the digital age. Institute of Library and Museum Services. Retrieved from /2014/11/indexing-interviews-in-ohms/.

Chicago

Boyd, Doug and Danielle Gabbard, Sara Price, and Alana Boltz. “Indexing Interviews in OHMS: An Overview,” in Oral History in the Digital Age, edited by Doug Boyd, Steve Cohen, Brad Rakerd, and Dean Rehberger. Washington, D.C.: Institute of Museum and Library Services, 2014, /2014/11/indexing-interviews-in-ohms/.

This is a production of the Oral History in the Digital Age Project (/) sponsored by the Institute of Museum and Library Services (IMLS).  Please consult /about/rights/ for information on rights, licensing, and citation.